Facilitating End-to-End Development of Individualized Therapeutics Public Workshop U.S. Food and Drug Administration Center for Biologics Evaluation and Research (CBER)
FDA White Oak Campus 10903 New Hampshire Ave. Bldg. 31 Conference Center Great Room Silver Spring, MD 20993
March 3, 2020
This transcript appears as received from the commercial transcribing service after inclusion of minor corrections to typographical and factual errors recommended by the DFO
2
ATTENDEES
Albert B. Seymour, PhD Homology Medicines, Inc.
Alison Bateman-House, PhD, MPH, MA NYU Grossman School of Medicine
Donald B. Kohn, MD UCLA
Guangping Gao, PhD University of Massachusetts
Jason J. Gill, PhD Texas A&M University
Jill A. Wood Phoenix Nest
J. Keith Joung, MD, PhD Massachusetts General Hospital
Malachi Griffith, PhD Washington University School of Medicine
Philip John (PJ) Brooks, PhD National Institutes of Health
Robert T. (Chip) Schooley, MD University of California San Diego Center for Biologics Evaluation and Research Gopa Raychaudhuri, PhD (CBER) Center for Biologics Evaluation and Research Peter Marks, MD, PhD (CBER) Center for Biologics Evaluation and Research Zenobia Taraporewala, PhD (CBER) Center for Biologics Evaluation and Research Roger Plaut, PhD (CBER) Center for Biologics Evaluation and Research Anita Richardson, MAS (CBER) Center for Biologics Evaluation and Research Sandhya Sanduja, PhD (CBER) Center for Biologics Evaluation and Research Zuben Sauna, PhD (CBER) Center for Biologics Evaluation and Research Rebecca Reindel, MD (CBER) Center for Biologics Evaluation and Research Larissa Lapteva, MD (CBER) Center for Biologics Evaluation and Research Zhenzhen Xu, PhD (CBER)
www.transcriptionetc.com
3
Center for Biologics Evaluation and Research Celia Witten, PhD, MD (CBER) Center for Biologics Evaluation and Research Capt. Julie Vaillancourt (CBER) Stephen Aldrich MyCancerDB
Joe Campbell NIAID
Matt Kelly Sarepta Therapeutics
Sharon Hesterlee Muscular Dystrophy Association Advanced Regenerative Manufacturing Richard McFarland Institute Karen Walker Genentech
Jessica Adomako Genentech
Marilyn Howard University of Pennsylvania Center for Biologics Evaluation and Research Carolyn Wilson, PhD (CBER) Lea Witkowsky Innovative Genomics Institute
Aron Stein Sangamo Therapeutics
Lorraine McLellan Audience Questioner
Lynne McGrath Audience Questioner
Neil Thakur ALS Association
Rich Horgan Cure Rare Disease
Anita Nosratieh FasterCures
Lauren Black Charles River Laboratories
www.transcriptionetc.com
4
TABLE OF CONTENTS
INTRODUCTION: DR. GOPA RAYCHAUDHURI ...... 6 SPEAKER: DR. PETER MARKS ...... 11 SESSION 1: MANUFACTURING ...... 22 SESSION 1 MODERATOR INTRODUCTION: DR. ZENOBIA TARAPOREWALA ...... 22 CHALLENGES AND OPPORTUNITIES IN DEVELOPMENT AND MANUFACTURING OF INDIVIDUALIZED THERAPEUTICS WITH AAV VECTOR-BASED GENE THERAPIES - DR. GUANGPING GAO ...... 30 DEVELOPMENT OF PHAGE THERAPY: PERSONALIZED MEDICINE AND INDIVIDUALIZED THERAPEUTICS - DR. JASON J. GILL ...... 51 PANEL SESSION WITH Q&A ...... 74 [BREAK] ...... 96 SESSION 2: TOOLS FOR SAFETY TESTING AND DEVELOPMENT ...... 96 SESSION 2 MODERATOR INTRODUCTION: DR. SANDHYA SANDUJA (CBER) ...... 96 PRECLINICAL APPROACHES/CHALLENGES IN DEVELOPMENT OF INDIVIDUALIZED THERAPEUTICS - DR. ALBERT B. SEYMOUR ...... 104 BIOINFORMATICS TOOLS FOR DEVELOPMENT, ANALYSIS & PRECLININCAL TESTING OF INDIVIDUALIZED THERAPEUTICS - DR. MALACHI GRIFFITH ...... 128 DEFINING OFF-TARGET EFFECTS OF GENE EDITING TECHNOLOGIES - DR. J. KEITH JOUNG ...... 150 PANEL DISCUSSION WITH Q&A ...... 174 SESSION 3: CLINICAL ...... 200 SESSION 3 MODERATOR INTRODUCTION: DR. REBECCA REINDEL ...... 201 OPPORTUNITIES AND CHALLENGES IN THE CLINICAL DEVELOPMENT OF BACTERIOPHAGE THERAPEUTICS - DR. ROBERT T. SCHOOLEY...... 209 CHALLENGES TO DEVELOPING INDIVIDUALIZED STEM CELL GENE THERAPIES - DR. DONALD B. KOHN ...... 230 PANEL SESSION WITH Q&A ...... 253 SESSION 4: PRODUCTS TO PATIENTS ...... 276 SESSION 4 MODERATOR INTRODUCTION: DR. CELIA WITTEN ...... 276 THE TRIALS AND TRIBULATIONS OF DRIVING A TREATMENT FOR AN UBER-RARE DISEASE TO THE CLINIC AND BEYOND-A PARENT’S PERSPECTIVE - MS. JILL A. WOOD ...... 286 ETHICAL ISSUES IN PRODUCT DEVELOPMENT AND SUSTAINABILITY FOR INDIVIDUALIZED THERAPIES-DR. ALISON BATEMAN-HOUSE ...... 308
www.transcriptionetc.com
5
BEYOND ‘ONE DISEASE AT A TIME:’ ACCELERATING CLINICAL TRIALS OF GENETIC THERAPIES BY GROUPING RARE DISEASE PATIENTS ACCORDING TO UNDERLYING DISEASE MECHANISM-DR. PHILIP J. BROOKS ...... 330 PANEL DISCUSSION WITH Q&A ...... 349 WRAP UP AND CLOSING REMARKS: DR. PETER MARKS ...... 378
www.transcriptionetc.com
6
1 INTRODUCTION: DR. GOPA RAYCHAUDHURI
2
3 DR. RAYCHAUDHURI: Good morning, everyone, and
4 welcome to the CBER workshop on facilitation and
5 development of individualized therapeutics. My name is
6 Gopa Raychaudhuri. I'm a senior scientist in the
7 Office of the Director at CBER, and I coordinate and
8 oversee CBER's individualized therapeutics program.
9 It is my distinct pleasure to welcome all the
10 stakeholders who are here today, including patients,
11 family members, patient advocacy organizations,
12 healthcare professionals, and individuals from
13 nonprofit organizations, academia, industry, and
14 government. In addition, I would like to thank
15 everyone who's participating in today's proceedings via
16 webcast. We appreciate you taking the time to join and
17 contribute online.
18 Before we begin today's proceedings, I have a
19 few general announcements. First, please silence your
20 cell phones and any mobile devices. If you haven't
21 done so already, we ask that all attendees sign in at
www.transcriptionetc.com
7
1 the registration tables just outside this meeting room.
2 If you would like to preorder lunch, you can
3 do so at the food kiosk that's just outside the
4 conference room. Lunch must be preordered by 9:30. If
5 you decide not to preorder, you may purchase snacks,
6 sandwiches, and other food items a la carte at the
7 kiosk. And that's open until 5:00 today.
8 This meeting is being transcribed, and a live
9 webcast is being recorded. There also is an official
10 photographer, as you can see, who will be taking photos
11 during the course of the workshop. For any urgent
12 issues, please speak to the registration staff that are
13 just outside or any FDA staff member with a tag, and we
14 will be able to assist.
15 I'd like to open today's workshop by defining
16 what we mean by individualized therapeutics and why we
17 need to think differently about development, licensure,
18 and access to these products. Out of thousands of
19 rare, hereditary, and acquired diseases, there are
20 hundreds of disorders affecting one or a small group of
21 people each year that could be addressed with
www.transcriptionetc.com
8
1 individualized or bespoke therapies. These therapies
2 are based on engineering a product aimed at the
3 specific mechanism underlying a patent's, or a small
4 group of patients', illness. That is the therapeutic
5 product is engineered specifically for a given patient
6 or small group of patients.
7 Examples include gene therapy or gene editing
8 for monogenic diseases, antisense oligonucleotides, or
9 genetically-engineered phages for multidrug-resistant
10 infection. Today's workshop will focus on
11 individualized therapeutic products regulated by CBER,
12 specifically, gene therapies and phage therapies. But
13 we work in close collaboration with our colleagues in
14 CDER, who are very active in this space to facilitate
15 development of antisense oligonucleotide products for
16 patients with rare diseases.
17 This slide shows the traditional regulatory
18 pathway, from discovery to licensure and post-licensure
19 monitoring, for biologics and other medical products
20 regulated by FDA. Individualized therapeutic products
21 present unique challenges because they do not fit the
www.transcriptionetc.com
9
1 traditional paradigm for manufacturing and clinical
2 development. This requires that we think outside the
3 box, to establish an evidence-based, clear, and
4 practical pathway for development, regulation, and
5 access for patients to these products, while assuring
6 that the standards for quality, safety, and efficacy
7 are maintained.
8 During the course of this workshop, you will
9 hear about existing challenges and potential solutions
10 to adapt the current process to meet the need and
11 opportunities for stakeholder collaboration to move
12 this field of work forward. This slide shows the large
13 number and wide range of stakeholders that play a
14 critical role in this field of work. As a community
15 working towards a common goal, each stakeholder brings
16 valuable perspective, knowledge, skills, and resources
17 to the effort. We are pleased that so many stakeholder
18 groups are participating in this workshop. And we look
19 forward to your contribution during today's
20 discussions.
21 The workshop is divided into four sessions.
www.transcriptionetc.com
10
1 The first is on manufacturing. The second session
2 focuses on nonclinical development and tools for safety
3 testing. The third is on clinical development. And I
4 should note that there is one change in the program in
5 the clinical section. Unfortunately, Dr. Kohn is not
6 able to be here in person because of flight delays
7 yesterday. But he has kindly agreed to join us
8 remotely and will be giving his presentation remotely.
9 And the fourth session will focus on what is the
10 ultimate objective, which is getting products to
11 patients in an efficient and sustainable way and the
12 critical role of partnerships and collaborations to
13 make this a reality.
14 Each session will start with a short
15 introduction by one of my FDA colleagues, who will
16 serve as the moderator for the session. That will be
17 followed by two or three 20-minute presentations by
18 external experts. And we are very happy to have each
19 of them here today.
20 Following the presentations, there will be a
21 panel discussion, where we will open the floor for
www.transcriptionetc.com
11
1 comments and questions for the speakers and panelists.
2 Online participants will have an opportunity to write
3 in their questions. This is an important part of
4 today's workshop because one of our main objectives is
5 to hear from you, the stakeholders, on what we can do
6 as individual organizations and collectively as a
7 community to facilitate end to end development and
8 timely and sustainable access to these important
9 products for the patients who need them.
10 And with that, I would like -- it's my
11 pleasure to introduce Dr. Peter Marks. Dr. Marks is
12 the Director for the Center for Biologics Evaluation
13 and Research at FDA. And he will give a few
14 introductory remarks on the current state, and he will
15 set the stage for today's discussion. Dr. Marks.
16 SPEAKER: DR. PETER MARKS
17
18 DR. MARKS: Thanks very much. Good morning.
19 And thanks everyone in the room for coming today. And
20 thank you all online for listening in.
21 I just want to spend a couple minutes setting
www.transcriptionetc.com
12
1 the stage for what leads us to be here today. And one
2 of those things is just to back up and say where we've
3 come over the past few decades. At the end of the last
4 millennium, we were in this era of personalized
5 medicine, where, increasingly, we were beginning to
6 understand characteristics of people's disease that led
7 us to choose medicines off the shelf that uniquely
8 address their diseases, whether that be for cancers,
9 infectious diseases, rheumatologic diseases.
10 And so this concept of personalized medicines,
11 where you find the right drug on the shelf to treat the
12 patient, became prevalent. But we're now moving into a
13 different era. And that's an era where we understand
14 the molecular defects at the level of the genome, and
15 we actually are needing to create the right drug to
16 treat the patient. We don't have things right off the
17 shelf. And so what we're really moving into is an era
18 of individualized medicine.
19 And yes, one could say that this is a semantic
20 difference, but there is a difference here in how we as
21 regulators, and I think how we as a field in general,
www.transcriptionetc.com
13
1 deal with this. And I just want to say that, as you
2 start to cogitate over this some, you realize that
3 individualized medicine actually breaks down further
4 into products that are essentially customized products
5 and products that are created. The customized products
6 are one where you could imagine they have the same
7 indication and the same mode of action, but there's
8 something different about each one that's unique for an
9 individual.
10 One can imagine that if one is making
11 dendritic cells against some cancer that a patient has,
12 some particular cancer -- it could be a lung cancer or
13 some other cancer. But if you have different product,
14 because it's got different peptides on it, you could
15 have a lot of different ones with some unique attribute
16 on some basic product. And those are customized.
17 On the other hand, there are products which
18 are going to be for different indications, different
19 lysosomal storage disorders, which may be very similar
20 in that they have the same backbone vector, if they're
21 a gene therapy, but they'll have different inserts.
www.transcriptionetc.com
14
1 And so I think this concept of customized and created
2 products is something that we'll hear more about today,
3 but it is something that we'll think about going
4 forward. And there probably are some regulatory
5 distinctions between these two classes.
6 Just to make this point a little further,
7 personalized medicine is a little bit like ready-to-
8 wear. You find stuff on the shelf that kind of fits
9 you, and it just is -- it's there. It's like medicines
10 off the shelf that you find that fit.
11 On the other hand, this concept of customized
12 products probably has the analogy of something that's
13 made to measure. Made to measure, for those of you who
14 are not familiar with tailoring -- and I'm not one to
15 talk because I'm not one for sartorial splendor -- made
16 to measure is actually not a custom-made suit, a made
17 to measure suit is one in which the fabric's been all
18 cut. It's just the seams are left open so that they
19 can be stitched closed once they adjust it to your fit.
20 And that's kind of the analogy to customized products.
21 On the other hand, there are these created
www.transcriptionetc.com
15
1 products, which is more like a bespoke suit. A bespoke
2 suit is somebody -- you choose the yards of cloth and
3 someone cuts them and puts them together. And so this
4 is kind of the analogy here I'd like to use. It's not
5 perfect, but it does give us the idea of what we're
6 dealing with.
7 Now, there are some challenges that go with
8 these individualized therapies. And that's what we're
9 here to talk about today. And we have sessions to talk
10 about manufacturing, nonclinical development, clinical
11 development, and product access. And we certainly
12 don't have the answers for these. But I think I'd like
13 to go through and kind of pose some of the questions.
14 Just to show you on the manufacturing side,
15 manufacturing in this area is really a challenge. In
16 the field of gene therapy, we have moved so fast in the
17 science we're now on to second and third generation of
18 genome editing constructs. We've moved on past the
19 initial generation of CRISPR to new novel genome
20 editors. That being said, we're still making vectors
21 much the same way that we made them at the turn of the
www.transcriptionetc.com
16
1 millennium. And we really need to figure out how to
2 move that forward.
3 The reason for that is that, right now, we
4 have one sweet spot, which is the grande-size cup in
5 the middle, where commercial viability relies on
6 products that are going to treat between a hundred and
7 maybe a thousand or a few thousand people. If one
8 wanted to treat very large numbers of patients --
9 that's something for another day's talk and another
10 day's meeting -- the technology is simply not there.
11 One would need essentially to fill Lake Erie as a
12 bioreactor to make enough vector to treat thousands of
13 people. That's just not where we are. We're going to
14 have to find more efficient ways to do that.
15 On the other hand, what we are here to talk
16 about today is how do we deal with making these
17 products for small numbers of individuals. And that's
18 a challenge because the setup costs currently to make a
19 gene therapy for 20 people are very similar to the
20 setup costs to make a gene therapy for 200 people, in
21 terms of a commercial process. So that's something we
www.transcriptionetc.com
17
1 have to really think about because it's limiting the
2 ability to get access to these products.
3 Nonclinical development is also a challenge
4 because animal models may be much less than ideal here.
5 And that's particularly true when we start to think of
6 genome editing because you might want to do toxicity
7 studies of a genome editor. You might want to
8 understand off-target effects on a genome editor.
9 But last I looked, although some people think
10 I look a little like a mouse, my genome does not look
11 exactly like a mouse. And it's a real problem.
12 Despite the fact that there's a fair amount of
13 identity, there's enough difference that you wouldn't
14 want to trust a mouse model for a genome editor with
15 the guide. So we may need to think about new model
16 systems, such as how can human organoid systems,
17 humanized mice, other models, help in this area for
18 safety testing.
19 For clinical development, I think we really
20 are going to have to think very novel-ly about how we
21 deal with very small populations. We're going to have
www.transcriptionetc.com
18
1 to think about how we document disease, the natural
2 history of disease, or collect baseline data so that,
3 when we actually have small numbers of patients
4 treated, we can see that there's some difference from
5 that baseline.
6 And once we start to treat patients, can we
7 find some way of using some type of continuous
8 reassessment model or some other statistical method, a
9 Bayesian or other design, to see if there's some point
10 at which one crosses where one declares victory that
11 you actually have an effective product, rather than
12 developing a sample size and running a traditional
13 clinical trial. And very importantly, we have to deal
14 with product access. How do we deal with the fact
15 that, for some of these product, once you've even
16 gotten them through development, how do you make sure
17 you have continued access to them given the complexity
18 and the cost of providing that access?
19 We already know what can happen. One of the
20 gene therapies that was approved in Europe is now no
21 longer on the market. Now, obviously, it can be made
www.transcriptionetc.com
19
1 available in other ways. But it shows you what can
2 happen here if we don't think about access and we don't
3 think about the costs and the complexity of these
4 products.
5 And so one of the things we've been thinking
6 about here is whether public-private partnerships could
7 at least help access in the United States and perhaps
8 even globally through the streamlined production of
9 products. Because when you think about it, if we're
10 making products for very small numbers of patients with
11 rare diseases, we shouldn't reinvent the wheel
12 globally. There shouldn't be initiative here in the
13 United States, initiative in Europe, initiative in
14 Asia. It probably would be better to have everyone
15 contributing to the effort.
16 So the idea here of a gene therapy public-
17 private partnership would be to try to drive down the
18 time it takes to get these products to patients by
19 using reproduceable processes or leveraging data to
20 make things happen. Because, typically, when
21 investigators, individual investigators develop a gene
www.transcriptionetc.com
20
1 therapy, they go through all the normal stages of drug
2 development, including submitting an IND, doing the
3 necessary tox studies, doing the manufacturing, which
4 oftentimes they have to learn to do from scratch. And
5 that whole process takes four to eight years often on
6 average.
7 One could imagine that if one was able to
8 leverage manufacturing processes, use templated INDs,
9 templated protocols -- obviously, it would have to be
10 customized somewhat -- you could potentially shorten
11 this a fair amount. And those -- that shortening of
12 years means a lot. For patients, when you talk to
13 parents of children with rare diseases, years matter, a
14 few years matter.
15 So our goal of this workshop is really to
16 think about where the opportunities are to adapt
17 processes from beginning to end. You see the
18 traditional process here. No one is suggesting that,
19 for these products, we're going to have traditional
20 phase one, two, three development.
21 The question is, where can we go with these to
www.transcriptionetc.com
21
1 end up at the end of the day with products that are
2 both safe and effective? Because I think, in the
3 process of trying to develop this paradigm, we want to
4 maintain the ability to know that, at the end of the
5 day, what we're providing to patients is something that
6 is similar in nature to the gold standard that we
7 provide patients with now with approved products. And
8 with that, I will turn it back over to Gopa.
9 But before I do, I just want to -- I just
10 nearly forgot one thing. I mean to have a slide here.
11 But I want to just take the moment to thank both Dr.
12 Raychaudhuri, as well as Leslie Haynes, for tremendous
13 work putting together this workshop.
14 There is a -- there are others as well from
15 our office who spent a lot of time taking care of
16 logistics of travel, making sure that things run well.
17 So thank you so much. And with that, I will wish
18 everyone a wonderful day. We look forward to the
19 discussions. Thank you.
20 DR. RAYCHAUDHURI: Thank you, Dr. Marks, for
21 that overview and vision of approaches and
www.transcriptionetc.com
22
1 opportunities that we can think about to advance
2 development and access to individualized therapeutic
3 products.
4 So we will begin the program now with Session
5 1. I would like to introduce Dr. Zenobia Taraporewala.
6 Dr. Taraporewala is a CMC reviewer, and Acting Team
7 Lead for the Gene Therapy Branch in the Division of
8 Cellular and Gene Therapies in the Office of Tissues
9 and Advanced Therapies at CBER. Dr. Taraporewala will
10 be the moderator for Session 1, which is on
11 manufacturing challenges and opportunities for gene
12 therapy and phage therapy products. Dr. Taraporewala?
13 SESSION 1: MANUFACTURING
14 SESSION 1 MODERATOR INTRODUCTION: DR. ZENOBIA
15 TARAPOREWALA
16 DR. TARAPOREWALA: Good morning. All right.
17 So let's get started. As Gopa said, I am Zenobia
18 Taraperwala, and I am from the Office of Tissues and
19 Advanced Therapies, commonly referred to as OTAT in
20 CBER. And I will be the moderator for the session.
21 We have two very distinguished speakers in the
www.transcriptionetc.com
23
1 session, Dr. Guangping Gao from University of
2 Massachusetts Medical School and Dr. Jason Gill from
3 Texas A&M University. Following the two presentations,
4 we will have a panel discussion. And the panel will
5 include speakers from this session and CBER
6 representatives from the Office of OVRR -- from OVRR
7 the Office of Vaccine Research Review, Dr. Roger Plaut,
8 and from the Office of OCBQ, which is Ms. Anita
9 Richardson. And during the panel, we will field
10 questions from the audience and from those attending
11 online.
12 So this is advancing but that is not. Okay.
13 Thank you. All right. So this slide illustrates the
14 manufacturing development, where manufacturing
15 feasibility is assessed, and analytical tests are -- so
16 this basically gives you a manufacturing development
17 paradigm that is currently followed for biologics and
18 drugs in general.
19 It starts with the discovery and the
20 preclinical stage where manufacturing feasibility is
21 assessed and analytical tests that are suitable for
www.transcriptionetc.com
24
1 release of research and developmental lots and IND-
2 enabling lots are developed. That manufacturing
3 experience is then leveraged for the manufacturing of
4 clinical lots under GMP suitable for early phase
5 studies. And with the release of multiple clinical
6 lots, there is greater understanding of the variables
7 in the starting and the raw materials and for the
8 process on analytical assays used in release testing,
9 all that are critical to ensure lot-to-lot consistency.
10 In late phase studies, the analyticals needed
11 for thorough product characterization and in-process
12 testing to meet the future demands of commercial
13 process are put in place. Finally, for licensure, the
14 expectation is that full GMPs are in place for
15 manufacturing of PPQ lots to demonstrate that the
16 process is capable of consistently manufacturing safe
17 and efficacious product at commercial scale. The
18 quality controls are in place at that time, and all the
19 assays are validated. The product stability during
20 storage and shipping has been demonstrated for
21 expiration labeling.
www.transcriptionetc.com
25
1 So considering this paradigm then, where are
2 the opportunities for flexibility in manufacturing for
3 individualized therapeutics? Next slide. When
4 discussing the challenge -- thank you. When discussing
5 the challenges to the development of individualized
6 therapeutics, one must consider the complexity of
7 manufacturing biologics for a large patient population
8 and whether the challenges apply or are similar for
9 individualized therapeutics, where one would expect
10 limited manufacturing experience with an N of one lot,
11 or patient-specific lots, or small lots.
12 What then are the challenges towards achieving
13 lot-to-lot consistency and quality for the licensure of
14 individualized therapeutics? So what do we recommend
15 sponsors do to ensure manufacturing consistency during
16 standard development of a biologic? It starts with
17 ensuring quality, safety, and consistency of the
18 starting material, minimize the risk of adventitious
19 agents, and ensure reliable supply chain.
20 The manufacturing process should be optimized
21 with robust process controls and a good understanding
www.transcriptionetc.com
26
1 of the variables, which comes with manufacturing
2 experience with the release of safe, potent, and
3 quality lots that are qualified or validated early in
4 development and generating well-characterized reference
5 standards that can be used throughout clinical
6 development. Consistency in product stability can be
7 achieved by developing a platform suite of assays for
8 quality control and release testing and by using assays
9 that are qualified and validated early in development
10 and generating well-characterized reference standards
11 that can be used throughout clinical development.
12 Product stability can be achieved by adopting standard
13 or previously tested formulations, standard container
14 closures, and storage conditions.
15 What are the considerations then to facilitate
16 the development of individualized therapeutics? Go to
17 the next. In this session, as we discuss and think
18 about the manufacturing of individualized therapeutics
19 and the related regulatory challenges, it is important
20 to note that regulatory flexibility is currently
21 afforded to patient-specific biologics and biologics
www.transcriptionetc.com
27
1 that are developed for rare and orphan diseases. And
2 such flexibility is afforded based on the risk/benefit
3 analysis on a case-by-case basis. For example, in
4 early-phase studies, assays that measure product
5 potency or processed impurities and release testing
6 plan for clinical lots may not be sufficiently
7 developed or qualified.
8 There is limited product characterization, and
9 we allow products that may have limited shelf life to
10 be released, based on rapid sterility testing and
11 sampling flexibility of small lots. In early-phase
12 studies, stability testing may be limited. Product
13 stability assessments are made on data collected in
14 real time from clinical lots that are placed on
15 stability.
16 This slide lists the examples of the
17 regulatory flexibility afforded in late-phase studies,
18 which includes release of clinical lots with potency
19 assay that measures the biological activity of product
20 that may not be completely -- an assay that is not
21 completely developed or qualified. We may allow for
www.transcriptionetc.com
28
1 release of lots with wide acceptance criteria for
2 release testing of some critical quality attributes of
3 the product. When limited lots are made during
4 clinical development, we have allowed comparability to
5 be established with limited data, for example, by a
6 side-by-side analysis of critical quality attributes
7 and without robust statistical analysis.
8 Process validation in support of licensure for
9 products that have limited demand, such as orphan
10 drugs, we have allowed PPQ protocols, which is process
11 performance qualification protocols that are developed
12 for process validation, and to release for process a
13 PPQ batch for the distribution before complete
14 execution of the process validation. So we call that
15 concurrent release. And that is consistent with the
16 FDA's guidance on process validation. It is key to
17 note that, even in such cases, any lot release
18 concurrently must comply with all CGMPs, regulatory
19 approval requirements, and PPQ protocol lot release
20 criteria.
21 So with that introduction to the session and
www.transcriptionetc.com
29
1 the manufacturing issues in general, we hope that the
2 speakers and the discussion following the presentation
3 will address the challenges of small-scale batch
4 manufacturing that may be needed for individualized
5 therapeutics, ways in which we can ensure manufacturing
6 consistency for such patient-specific therapies, the
7 regulatory challenges in manufacturing and how these
8 can be addressed, and whether there is a need for
9 additional guidance, standards, or other regulatory
10 tools. So with that introduction to the session, it is
11 my pleasure to invite Dr. Guangping Gao, who is the co-
12 director of the Li Weibo Institute for Rare Disease
13 Research. He's also the director of the Horae Gene
14 Therapy Center and Viral Vector Core.
15 He holds multiple appointments at the
16 University of Massachusetts Medical School and is an
17 elected fellow for the U.S. National Academy of
18 Inventors and the American Academy of Microbiology.
19 He's also the president of the American Society of Gene
20 and Cell Therapy. He's an internationally recognized
21 gene therapy researcher. He's been instrumental in the
www.transcriptionetc.com
30
1 discovery and characterization of many AAV serotypes
2 that have revitalized the gene therapy field. And he
3 has made several impactful contributions in this field
4 of AAV vector gene therapy.
5 With that introduction, I welcome you to give
6 the first talk.
7 CHALLENGES AND OPPORTUNITIES IN DEVELOPMENT AND
8 MANUFACTURING OF INDIVIDUALIZED THERAPEUTICS WITH AAV
9 VECTOR-BASED GENE THERAPIES - DR. GUANGPING GAO
10
11 DR. GAO: Thank you to FDA and our colleagues
12 here inviting me to talk about AAV gene therapy for
13 individualized therapeutics. So what I would like to
14 use this time to tell you about some basic concepts and
15 challenges in AAV gene therapy for individualized
16 medicine. This is my disclosure.
17 And so gene therapy strategy, I would like
18 first to take a broad view with you what are the gene
19 therapy approach strategies. The first one is
20 traditional or conventional medicine developments, that
21 is you use gene therapy such as AAV based vector as a
www.transcriptionetc.com
31
1 drug directly given to human patients and accomplished
2 therapeutic efficacy. And here, we have three drugs
3 approved by EMA and by FDA.
4 The second strategy is, basically, genetically
5 modify the cell therapy. In this case, you take out
6 human cells and genetically modify in vitro, expand it
7 and give back as a live medicine, a living medicine,
8 and give back to patients. Okay. And there are three
9 drugs approved by EMA and FDA.
10 So to accomplish gene therapy, I personally
11 believe there are four critical components: therapeutic
12 gene and the pathomechanisms, vector delivery methods,
13 and animal model to perform preclinical product
14 developments. And among those, the delivery vehicle
15 vector is most important. So if you look at the field
16 of the gene therapy in the past two-and-a-half decades,
17 you can see that, basically, gene therapy has been
18 going through different vector platform, includes
19 adenovirus and lentivirus.
20 So we, as the gene therapists, our wish list
21 is the following. We want the vector to have a high
www.transcriptionetc.com
32
1 efficiency, long-term stability, low immunogenicity and
2 toxicity, and no genotoxicity. So among those
3 features, adeno-associated virus has it all. So this
4 virus is really a teamwork inside-out, with the capsid
5 determining where the vector should go and how to get
6 there, where to drop the gene payload in the right zip
7 code.
8 And the vector genome itself, it actually
9 carries gene payload for gene therapy. One of the
10 major limitations that we have here is adeno-associated
11 virus is very small. It only can carry the transgene
12 cassette, no more than 4.6, 4.7 kb.
13 And however, this genome is critical in terms
14 of formed, stable episome vector structures to be
15 persistent in terminally-differentiated tissues. So
16 how we transform this almost a virus -- because without
17 helper, this virus is doing nothing -- into amazing
18 vector. So basically, what we do is we take out wild-
19 type genome section of regulatory protein and a capsid
20 protein. And then we put back our expression cassette,
21 therapeutic gene. And then we provide helper function
www.transcriptionetc.com
33
1 in trans. We provide the packaging plasmid, the
2 regulatory sequence, and capsid sequence in trans.
3 If we put those three things together, either
4 by transient transfection, stable transfection, or
5 infection, we can create -- get into the producer cells
6 and then generate AAV vectors. The beauty of this
7 virus is that what AAV dressed doesn't matter. So what
8 you can do simply is change the dress, change the
9 capsid and become a new vector. For example, this we
10 change to AAV8. 9 capsid become AAV9 vector. And then
11 if you change to AAV8, it becomes AAV8 vector.
12 So exactly how AAV works, I just want to use a
13 simple cartoon to demonstrate how AAV works as a gene
14 therapy drug. Basically, like any other viruses,
15 AAV’s getting to the cells by receptorology process,
16 and then going through endosome trafficking. There you
17 have two pathways. One, you probably can get into a
18 proteasome, get degraded. And another, you can enter
19 the nucleus. And efficiency or the ratio of those two
20 proportions of the viruses depends on serotypes.
21 Some serotypes have more efficient escape from
www.transcriptionetc.com
34
1 endosome, and some it's more efficient to get into the
2 cells, into the nucleus. Once you get into the
3 nucleus, the first step AAV need to uncoating and then
4 that release, let single strand AAV genome are both
5 negative strand and positive strand. Then you can form
6 double-strand genome through host machinery or self-
7 annealing. And then, you generate a vector genome that
8 is a ready to expression.
9 However, there's a mechanism because of a
10 terminal repeat of AAV8. It can form a circular
11 version and then can be stabilized in a terminally-
12 differentiated tissue. Very small portion of AAV could
13 be potentially integrated.
14 And then once you have a stabilized genome
15 that you have a gene expression, you can generate gene
16 therapy products in this case. So if you look at the
17 AAV gene therapy -- and currently, AAV gene therapy has
18 entered an exponential growth period. And there are
19 nine different therapeutic areas using AAV gene therapy
20 vector.
21 Ophthalmology or ocular gene therapy and CNS
www.transcriptionetc.com
35
1 gene therapy takes the lead but followed by many other
2 different applications. So now we know -- we talk
3 about this strategy. We talk about the vector
4 delivery. And exactly how can you accomplish gene-
5 based therapy, I summarize here in four different
6 formats.
7 The first one is a classic gene therapy, that
8 when you have a gene not functional, you can provide a
9 healthier gene to replace these malfunctioning genes.
10 This is a very classic version of the gene therapy.
11 The second part is that you may have nothing wrong with
12 your gene, but you need certain gene products to fight
13 diseases such as cancer or infectious diseases. You
14 can add genes to fight those diseases.
15 The third one is, when you have a genetic
16 mutation, when you get a gain of toxicity, gain of
17 function outcome, you can silence this gene and
18 accomplish therapeutic treatment. And the final one
19 could be gene editing. And this is now entering a
20 rapid growth using gene editing as a tool for gene
21 therapy.
www.transcriptionetc.com
36
1 So now, I would like to use one of my
2 personal-professional journey about a disease called
3 the Canavan disease demonstrating how gene therapy
4 works. This disease was discovered in 1931 by Dr.
5 Canavan. And this is a very tragic disease,
6 devastating disease.
7 Basically, you have a quite high prevalence in
8 Ashkenazi Jewish population, but also, in general
9 population, you have 1 in 300 mutations. The genes
10 could've been mutated. And then this patient dies
11 usually before five. There's no medicine to treat this
12 patient -- those patients. And this is a spongy
13 degeneration, the disease is.
14 And this is a picture I took in 2013, when I
15 went to a Canavan disease meeting and I took this with
16 a dozen of the patients. And 2016, I went back to the
17 same meeting, only a few kids still running around.
18 And this really picture remind me, we as gene therapy
19 researchers, we have to work as hard as we can to have
20 the race with time to develop a therapy to save the
21 lives of those children.
www.transcriptionetc.com
37
1 And as you can see, because of this
2 leukodystrophy is a spongy degeneration, the brain,
3 whole brain is just like Swiss cheese. And this is my
4 mentor, PhD mentor, Dr. Rueben Matalon, who discovered
5 biochemical defect of this disease in 1988. Basically,
6 what happened, there's a product called NAA made in
7 neurons, in mitochondria. And they transport into
8 oligodendrocytes where it should be degraded by
9 aspartoacylase.
10 However, if we have this gene mutation, then
11 all the tragic starts here. You have NAA accumulating
12 in inter-tissue space in the brain, and then you
13 generate spongy degeneration. I entered this Canavan
14 research in 1989 as a graduate student and published my
15 thesis research in 1993. And this allows the gene
16 therapy possible. But the issue is, as everybody knows
17 that -- the key issue, as I said from the beginning,
18 it's the vector.
19 So I decide to start my post-training with Dr.
20 Jim Wilson at the University of Pennsylvania. And
21 then, Jim and I started this joint venture that is
www.transcriptionetc.com
38
1 looking for more efficient AAV. And we basically
2 designed primers across the variable regions of the AAV
3 genome, amplifying the capsid. Using this method, we
4 generated a library of AAV capsids.
5 And I want to give you an example. This is
6 one of the most expensive but most popular virus,
7 that's AAV9 -- everybody knows that, AAV9, so widely
8 used for different applications in gene therapy. So we
9 isolated this from human tissue, human liver. And
10 that's the original lab record showing what isolates
11 clone 28.4 as a gene AAV9 vector. And the beauty of
12 this vector is it can cross blood/brain barrier and
13 transduce the brain quite efficient.
14 So with this available, then we, at the
15 University of Massachusetts, start to develop gene
16 therapy for the disease. This is my MD PhD fellow, Dr.
17 Dominic Gessler. So basically, we developed it using
18 AAV9. And through extensive optimization, we developed
19 the vector treat an animal model exactly recalculate
20 human disease.
21 As you can see, the gene therapy here,
www.transcriptionetc.com
39
1 basically, this is Canavan disease. The neurotrack,
2 it's very short. This is normal. But after gene
3 therapy, you basically, through this tractography, you
4 see we restore neurotrack in the patient mice.
5 And very importantly, this gene therapy can
6 reverse pathology. As shown in here after one week of
7 gene therapy and this Canavan disease still there. And
8 this is wild-type, and this is Canavan treated there,
9 so not yet. But if you look at week four, then the
10 situation will be very different. And this is the
11 Canavan disease that we have treatment, this is normal,
12 this is treated. As you can see, we basically restored
13 myelination and reversed pathology.
14 So with this available -- and I collaborated
15 with Dr. Barry Byrne of the University of Florida -- we
16 started as a Phase 1 -- not Phase 1, sorry -- expanded
17 access trial with two-year-old patients. And
18 basically, we -- in this process, we also applied
19 immunosuppression to leave a potential window of
20 treatment for future dosing if this treatment didn’t
21 work. And as you can see, treatment is safe.
www.transcriptionetc.com
40
1 Transaminitis, we do not have transaminitis elevation.
2 And also, this patient growth curve, as showing here,
3 is quite encouraging.
4 And so if you summarize the data from this
5 expanded access, is we know the immune modulation
6 worked perfectly because up to today, we still cannot
7 detect any anti-AAV9 antibody in the patient's
8 bloodstream.
9 And also, AAV gene therapy improved the
10 myelination and also the motor development and restored
11 the vision. The patient could not see before. Now,
12 when you walk in the room, the patient can follow you
13 around with his eyes.
14 And also, using DTI, the MRI, we demonstrate
15 that we show the development findings as we saw the
16 patients start moving with their hands. And this is a
17 patient before the treatment, and this is the patient
18 about, I would say, 15 months after treatment. You can
19 see, really, we improved their mobility and also
20 quality of life.
21 And whenever I saw these pictures, I feel very
www.transcriptionetc.com
41
1 much rewarding as a scientist who spends your life and
2 your professional research on this disease. The family
3 will be able to see the patient can save his life and
4 improve the quality of his life. So this is a powerful
5 -- a powerful outcome of the gene therapy. However, I
6 want to remind you, with gene therapy, we still have a
7 long way to go.
8 And in this journey of gene therapy express
9 reach to hospital usage and clinical application, we
10 have several hurdles we're going to face. The first
11 one, number one, I can see is the manufacturing. And
12 we -- I don't think we have enough vector to treat
13 patients as a commercial drug. And particularly, the
14 potency of the vector, we also need to improve.
15 The second part is this pre-existing and also
16 adaptive capsid transgene responses. And this can --
17 immunotoxicity could generate safety concerns. The
18 third part is, even though we want to say we should
19 have cell-specific, tissue-specific gene therapy -- but
20 I have to tell you, at this stage, my personal belief
21 is we only have efficient vector and nonefficient
www.transcriptionetc.com
42
1 vector. I don't think we really can control tissue
2 tropism and cell specificity.
3 And the final hurdle is, so far, we thought
4 the gene therapy, the more the better. Actually, soon,
5 I personally believe we will see that we need to have a
6 regulated gene therapy, long-time, overexpressed, super
7 physiological level may not be good. So once we have
8 those hurdles overcome, we'll have regulated gene
9 therapy arrow, have as a drug for clinical usage.
10 So how to address those questions? I want to
11 give you one example. That is we continue looking for
12 better vectors and for -- basically, this is an effort
13 we have done -- and generate about -- so through the
14 PCR and the high-throughput sequencing, we generate
15 about 70,000 more AAV capsid. Through complicated
16 bioinformatics pipeline, we eventually identify about
17 1,000 vectors. And some of those vectors can overcome
18 the production hurdle. AAV2 is one of the poorest
19 producers, as you guys doing gene therapy should know.
20 But we found in those variants, AAV2, 25
21 percent of them produce much better than AAV2. Some of
www.transcriptionetc.com
43
1 them produce 20-fold higher than AA2. And I want to
2 show you one example of such efforts. And that is we
3 isolated AAV2 variants, only have 13 amino acid
4 difference from AAV2.
5 However, this translates to -- into 13-fold
6 more efficient than gene delivery, seeing as you can
7 see here, you have AAV2 injected in one side of
8 hippocampus. And then if you look at other side
9 injected with new vectors, you can see 13-fold of
10 enhancement. It tells you we still have the room to
11 improve our vector.
12 And we also continue this kind of search. The
13 first one, we screened about 38 vector, and then here,
14 we screened about 50 vector or so. We identified 21
15 vector that's as efficient as AAV9. Seven of them, I
16 actually can see either equal or pass AAV9 performance.
17 So it tells you we have further room to improve
18 identify more efficient vectors for CNS gene therapy.
19 So -- but now, for gene therapy manufacturing,
20 that's one of the topics of today's meeting. And so
21 far, the current platform technology, there are two
www.transcriptionetc.com
44
1 categories.
2 One is transfection-based. This is developed
3 by Drs. Xiao Xiao and Jude Samulski in 1998. This is a
4 very popular process for AAV manufacturing. And then
5 if you look at the other category that's infection
6 based, either HeLa -- producer cell line HeLa, rep/cap
7 cell line, herpes virus, or Sf9 Baculovirus. Those are
8 infection-based process.
9 But if you look at the current -- I give a
10 snapshot of where we are in terms of manufacturing --
11 you can see here 293 transfections still dominate the
12 manufacturing process. And second is a Baclovirus
13 quickly catch up. And third on is a HeLa-based -- the
14 herpes virus or adenovirus-based and herpes virus-
15 based. So this is kind of current trend in
16 manufacturing platform technology. However, if you
17 look at the whole field, you will see CDMO's play a
18 major role, only about 30 percent of the company, doing
19 AAV themselves, and CDMO becomes a rapid-growth
20 industry here.
21 So now, for AAV manufacturing, what are the
www.transcriptionetc.com
45
1 major challenges we are facing? So I would say it
2 depends on many different factors, such as targeted
3 tissues, eyes, versus muscle, and route of
4 administration, subretinal injection versus systemic
5 injection and patient population size or patient size,
6 such as pediatric versus adult patients, and the
7 serotype and transgene cassette also contributed to
8 manufacturing. But the gap between the current
9 producibility and the clinical needs is either a one-
10 log, two-log, or three-logs, depends how big the
11 patient population size is.
12 So the second issue is how are we going to
13 close this gap? I personally believe we need -- really
14 cannot just do by increasing the bioreactor size, and
15 we need to enhance yield per cell. This will
16 definitely require collaborations between vector
17 biologists and bioengineers.
18 So one thing we should consider is AAV
19 replication and packaging biology, understand there's a
20 timing sequencing and a level of different gene
21 expression. The second part is cellular factor. We
www.transcriptionetc.com
46
1 have to figure out the cellular factor that can further
2 boost replication of packaging. And third is, ideally,
3 we should have a producer cell line, real producer cell
4 line, just adding the compounds that trigger entire AAV
5 manufacturing.
6 And this we have a lot of work to do ahead.
7 Of course, like any other FDA-approved product, quality
8 control and the bioanalytics is very important. One of
9 the major issues with AAV is empty-to-full ratios. And
10 right now, because it's manufacturing we’ve still not
11 optimized yet, it’s 90 percent empty to 10 percent
12 full, and if you do not have a special process to
13 purify it -- but our goal is to reverse it, 10 percent
14 empty and 90 percent full.
15 The second part is a biopotency assay,
16 infectivity assay. This is very hard because, ideally,
17 you should develop a kind of across different product
18 pipeline, infective assay. And we also need a
19 biopotency assay. This is product dependent.
20 And finally, it's about the vector genome
21 itself. And currently, for many different reasons, we
www.transcriptionetc.com
47
1 really do not understand what we have in AAV package.
2 I'm going to show you an example here. So if you put
3 guide RNA into this construct, if there's a simple
4 guide RNA structure, you can see you get a uniform
5 band, single band. However, what's exactly in this
6 band, we do not know.
7 My colleague, Phil Tai start looking to this
8 black box of AAV. He developed this sequencing for ITR
9 to ITR, every single molecule of AAV prep and to
10 understand exactly what's there. You can see the
11 majority of the vector here is full-length.
12 And however, if you put two guide RNA into the
13 vector, and then in these head-to-tail -- head-to-head
14 configurations, you can see you have many truncated
15 genomes. And if you use the sequencing we call the
16 AAV-Gp Seq we consider as a next-generation QC
17 pipeline, you can see you virtually have no full-length
18 molecule. You have many different truncated forms of
19 the molecule. If without this technology, you do not
20 know exactly what you have.
21 And however, interestingly, if you change that
www.transcriptionetc.com
48
1 configuration to head-to-tail of the two guide RNA,
2 then you come predominantly a single band. And this is
3 confirmed by sequencing itself. So this tells us we
4 still have a way -- a lot of work to do to figure out
5 quality control and give the best medicine, safest
6 medicine to our patients.
7 And so finally, I want to summarize what's the
8 difference between this individualized therapy-based
9 traditional gene therapy. And so we have one problem.
10 We have realized that, pre-clinical development, we
11 need better understanding the genetic causes and the
12 pathomechanism.
13 Depends on -- it’s loss of function, gain of
14 function, autosomal recessive versus dominant, X-link
15 or haploid insufficiency, this -- you will have a
16 different strategy for gene therapy. And also, many
17 ultra-rare diseases, there's no animal model for it.
18 So this in vitro versus in vivo model versus normal
19 versus disease model, expressing versus the function
20 and versus the phenotype correction, those are the
21 things we have to figure out at least, those ultra-rare
www.transcriptionetc.com
49
1 diseases with the patient of 1 to 10 to 100.
2 And also, the timing and the cost remain the
3 same. If you do a gene therapy drug to preclinically
4 develop this drug, it's the same cost, the same time.
5 It doesn't matter it's 10 patients or 10,000 patients.
6 And I don't think from NIH funding point of view, when
7 you -- disease from a small -- I don't know how much
8 funding opportunities we could get.
9 And the regulatory and supportive flexibility,
10 again, to generate this data to support in vivo data
11 for efficacy to help prepare IND and prepare the
12 pivotal toxicology studies, I think the time and
13 regulatory support required is same. Finally, GMP
14 manufacturing and the QC bioanalytics, so most of the
15 challenge remains the same. It doesn’t matter this
16 drug for 1 patient, 10 patients, or 10,000 patients.
17 And additional unique challenges include technology
18 transfer and the process development lot cost remain
19 the same. It doesn't matter it's 1 lot, 1,000
20 patients, or 10 patients.
21 And cost for the smaller batches may not be
www.transcriptionetc.com
50
1 even linearly scaled down. You probably cost more for
2 small-pack batches. And 1 dose versus 100 doses, if we
3 have existing patient treated with 10 patients treated,
4 what about a future patient? Where we should store and
5 maintain the stability and distribution of those gene
6 therapy vectors? I think as that's some of those
7 topics we should discuss after this meeting. Thank you
8 very much.
9 DR. TARAPERWALA: All right. Thank you, Dr.
10 Gao. Our next speaker is going to be Dr. Jason Gill
11 from Texas A&M. He was born and raised in Canada and
12 received his BSc and MSc degrees from Brock University
13 and his PhD in Food Microbiology from the University of
14 Guelph in Canada.
15 He holds a faculty position at the Department
16 of Animal Sciences at Texas A&M University, where the
17 major research focus in his lab is the biology and
18 application of bacteriophages. These are often also
19 referred to as phages. And you're going to hear more
20 about it. These are viruses that can infect bacteria.
21 Specifically, the research in his lab includes
www.transcriptionetc.com
51
1 phage genomics, basic phage biology, and application of
2 phages in real-world settings against many bacterial
3 pathogens. Dr. Gill holds many joint appointments in
4 the faculty of genetics in the Department of Microbial
5 Pathogenesis and Immunology in the College of Medicine.
6 And he serves as the associate director for the Center
7 for Phage Technology and Interdisciplinary Research and
8 Teaching Initiative, supported by Texas A&M AgriLife
9 Research. So welcome, Dr. Jason Gill.
10 DEVELOPMENT OF PHAGE THERAPY: PERSONALIZED MEDICINE AND
11 INDIVIDUALIZED THERAPEUTICS - DR. JASON J. GILL
12 DR. GILL: So good morning. I'd like to thank
13 the organizers for inviting me here to give this talk.
14 So what I'm going to talk about -- what I was asked to
15 talk about development of phages, as a therapeutic and
16 really, more on the manufacturing side. And so to
17 understand the manufacturing part, we're going to have
18 to understand more about the biology of that as well.
19 This is my COI statement. So I did some consulting for
20 Merck last year.
21 So to understand about the development of
www.transcriptionetc.com
52
1 phages as a therapeutic, we have to understand that
2 phages are very diverse. And so Dr. Gao talked about
3 adenoviral vectors. And they're quite variable. And
4 so phages are probably the most diverse set of
5 organisms on earth. They're very ancient, and they've
6 been coevolving with bacteria for many billions of
7 years.
8 They tend to be very adaptive to their hosts
9 at the strain and species level. So they're very
10 specific. They have very diverse genome content. So
11 you'd have two phages that infect the same host which
12 have really zero detectable DNA sequence similarity at
13 all. So there's a large diversity there in terms of --
14 that drive how you do your developments. And there's
15 also a lot we don’t understand about how phages work on
16 a more fundamental level.
17 So a lot of phages carry like hypothetical
18 genes, genes of unknown function, which as far as we
19 can tell have no phenotype. But again, for regulatory
20 purposes, that may become more important. So the
21 development of phages, this is actually not a new idea.
www.transcriptionetc.com
53
1 So phages were actually used as a relatively mainstream
2 therapeutic up until the 1940s when antibiotics were
3 developed. And then they were kind of abandoned in the
4 west and revived really with the rise in antimicrobial
5 resistance, starting really intensely about 20 years
6 ago.
7 So this is from an older paper. But this lays
8 out two theoretical development pipelines for using
9 phages. So the top one, you have this -- what's called
10 “pret a porter” system. And this is what was talked
11 about in the opening talk with this idea that you kind
12 of have a mass-produced predefined product. So you'd
13 isolate a phage or a set of phages against a particular
14 pathogen. You would understand them as well as you
15 could. You would have a GMP manufacturing process.
16 You take them through a regular phase one,
17 two, three approval pipeline. And then, you would mass
18 produce that product and then market it. You'd have
19 like a CGMP-produced, very well-defined product.
20 And the alternate approach on the lower panel,
21 in blue, is this kind of “sur mesure” idea, which is
www.transcriptionetc.com
54
1 this made to order idea. And so this is that you would
2 have a collection of phages, maybe 100 phages, maybe
3 200 phages, 500 phages, in a library. You would get
4 bacteriology back from a patient.
5 You would then identify phages that worked
6 against that particular -- that patient's particular
7 strain because remember phages are very specific for
8 their hosts. And you would then work that up into some
9 kind of product then administer that to the patient.
10 So the idea there is that when you have this
11 kind of product, you have tying to the bacteriology.
12 But you have a lot of issues on the regulatory side.
13 So the product identity is not as clear in that kind of
14 situation. The manufacturing is likely to be less
15 controlled. And this really thought of -- this is
16 about a 10-year-old paper. And this was thought of as
17 really kind of a -- not necessarily with the idea of
18 developing a large commercial product in mind. This is
19 really a way to treat patients.
20 And so this kind of approach works well if you
21 want to just kind of keep doing like expanded access or
www.transcriptionetc.com
55
1 eIND administrations, which have been happening. But
2 it's hard to see how that would be commercializable.
3 So I think what we're talking about here today is this
4 personalized medicine approach, where you would want to
5 take part of that sur mesure idea, where you have this
6 library of phages, and try to then take that through
7 some kind of regulatory process. So that could
8 actually be a widely available, possibly even mass-
9 produced product.
10 So product identity is pretty important for
11 any kind of therapeutic. And so on the -- if you have
12 a fixed product, that's a lot easier to define. So
13 you'll have a phage isolate or a set of phage isolates.
14 They'll be grown on a host or a set of hosts. There
15 will be some kind of production and purification
16 process that will be defined. And then you end up with
17 a product at the end. So that's a lot more straight-
18 forward, even for a biological.
19 And personalized approaches becomes -- the
20 idea of product identity becomes a little fuzzier. So
21 if you have, say, a collection of phage isolates, they
www.transcriptionetc.com
56
1 may or may not be grown on the same host. So you're
2 going to have that variance. You're going to have not
3 just new phages but phage host pairs in the
4 manufacturing process. There's going to be possibly
5 different manufacturing processes for different phages.
6 They may not all work in exactly the same manufacturing
7 process.
8 And then you're going to end up with,
9 hopefully, some CGMP manufactured individual phage
10 isolates, which you then blend together and then give
11 to a patient. And then so when you're talking about
12 product identity -- so is the product identity each of
13 the individual phages? Is it the final mixture of
14 phages? And right now, I don't know how -- if that
15 aspect is very clear.
16 So the size of the library you're going to
17 need is also going to be pretty variable between things
18 you want to treat. So different pathogens have
19 different levels of diversity, and that will affect the
20 size of the library that you're going to need to have.
21 So if you have very diverse targets, like for example,
www.transcriptionetc.com
57
1 Acinetobacter baumannii or Burkholderia cenocepacia,
2 which are organisms that we deal with, the host range
3 of those phages is really narrow.
4 So on the right, there's some data from our
5 lab on some collections of phages against Acinetobacter
6 baumannii. And this isn't the full table; this is just
7 the table that shows where the hits are. There's
8 another panel about this size, where the phages also
9 don't infect any of those strains. So the host strains
10 with these phages are very narrow. But if you look at
11 organisms like, for example, staph aureus or listeria
12 monocytogenes or shigella, those organisms are much
13 more clonal. And in that case, you can have a smaller
14 phage collection, which will cover most or more of
15 those strains.
16 So the target really affects the diversity of
17 the library. So if you're talking about a personalized
18 approach, if you're talking about this for staph
19 aureus, you may need a library of only 20 phages or 10
20 phages. But if you want to do this for Acinetobacter
21 baumannii, you're probably going to need a few hundred
www.transcriptionetc.com
58
1 on hand if you want to actually be able to cover most
2 strains of the pathogen.
3 So phages can be developed in a few different
4 ways. So they can be completely natural. They can
5 contain engineered components. This is -- some of this
6 also is still really evolving in the field. So you can
7 have the engineering be present in the phage
8 chromosomes. Actually, when the phage replicates, that
9 is a transmissible change, or they could also be
10 expressed in trans. So really, just the phages you
11 produced have the engineered or modified component, but
12 they don't pass that on to their offspring. Or they
13 may not produce any offspring at all, depending on how
14 you've engineered them.
15 There's a few companies that are looking at
16 synthetic phages right now. But you have to keep in
17 mind, even when they're talking about a synthetic
18 phage, it's still based on natural DNA sequence from
19 other phages. It's just been pieced together. So
20 there's a few different ways then that these are being
21 developed. There's a replicative phage therapy, which
www.transcriptionetc.com
59
1 I call this the classic approach.
2 So this is you're taking natural phages from
3 the environment, selecting them, working them up into
4 some kind of treatment, and then they are going to
5 infect cells. And they're going to replicate and lyse
6 those cells and release progeny. And that can happen
7 with naturally occurring phages or with engineered
8 phages.
9 And another route which is being explored is
10 using highly engineered phages, which don't necessarily
11 replicate. And they will -- they might just adsorb to
12 a cell and then deliver some kind of cytolethal agent,
13 like a CRISPR protein or something else that would
14 cause the cell to die. Or it's possible to have a
15 phage deliver a genetic payload to a cell to alter its
16 phenotype, for example, to just remove resistance
17 plasmid. Or it's even been proposed to say allow the
18 cell to now start synthesizing new metabolites and that
19 -- this is like a microbiome engineering type of thing.
20 But these are various ways that you can --
21 they are neat phage tricks that you can do. So they're
www.transcriptionetc.com
60
1 quite versatile in how you can engineer them. And this
2 isn't exhaustive either. I'm sure people are thinking
3 of all kinds of wild new ways you can use phages.
4 So one of the limitations here is the
5 knowledge base of phage biology as it exists today. So
6 phages are relatively complex viruses. This is a map
7 of the genome of phage lambda. And you can argue that
8 phage lambda is the best understood organism in biology
9 because of its long history of study. It has about a
10 49-and-a-half kb genome, as you can see here. But even
11 though this phage has been studied for, like, thousands
12 of person years of work likely, there are still some
13 genes we actually -- the functions are not known.
14 So if you can delete those genes, they don't
15 appear to have a phenotype. And those are highlighted
16 there in green. The rest of it is all pretty well
17 understood.
18 But then we move that focus from lambda. This
19 is what we call an inch wide and a mile deep. So we
20 know a lot about lambda. If you move that focus over
21 just a little bit, this is a group of phages that are
www.transcriptionetc.com
61
1 morphologically similar to lambda.
2 They infect salmonella. They're called 9NA-
3 like. 9NA is the first of this type to be isolated, so
4 that's what they're all called. So they have a
5 relationship to lambda. They're roughly the same size,
6 a little over 50kb. But if we look at this group of
7 phages, there's 176 genes in this group of phages.
8 Only 32 of those genes actually have a
9 predicted function. And 143 don't have a known
10 function, at least not what we can tell
11 bioinformatically. And 68 don't even have a
12 counterpart in the database. So they're completely
13 novel genes.
14 And this is not that unusual when you're
15 finding new phages against new bacterial targets that
16 you'll have a section of protein-coating genes.
17 Actually, they're the first example of those genes that
18 have been found in biology. So we've done this for
19 Caulobacter. And in that case, I think something like
20 two-thirds of the genes were actually novel in those
21 organisms.
www.transcriptionetc.com
62
1 So phage production is going to be based
2 around -- it's a biological manufacturing process. So
3 they're generally going to have to be propagated in a
4 live host. It's not very hard to get titers of 10 to
5 the 9 or 10 to the 10 pfu per ml in a culture. That's
6 just in the lab. If you optimize that, you can
7 probably drive that up to 10 to the 11 or maybe 10 to
8 the 12 even.
9 There's a few different ways you can get from
10 your crude production lysate down to the final product,
11 which are up here. So you're starting with a lysate to
12 somehow -- and you can use a few different methods to
13 take you from that, which is going to be the crude
14 liquor, basically, of the culture, down to somewhat of
15 a clarified lysate. And then there's a few
16 technologies that exist to take you from that to kind
17 of a relatively pure concentrate, which you can then
18 put through additional polishing steps, too, so we can
19 help remove endotoxin or further purify it.
20 So the clarified lysate is really just going
21 to be spent cultured media with phage in it. It's
www.transcriptionetc.com
63
1 going to have a lot of cell debris and media components
2 in it. A clean concentrate might be relatively pure.
3 It's mostly phage. And it will be more concentrated.
4 Most of the media components have been removed
5 and the cell debris. The endotoxin level will be
6 reduced. It may not be as low as you need, so you may
7 need to do additional steps to remove endotoxin or to
8 remove additional contaminants until you get to a
9 purified phage, which you would then blend and package.
10 So the methods are still very much in
11 development. And the different companies have their
12 own internal processes for that. But they're not
13 completely alien, I don't think, to the world of
14 manufacturing. This is a -- they're not that
15 different, say, from production of common proteins from
16 bacteria or certain viral vectors or vaccines. You're
17 having a biological production process. The
18 technologies aren't -- they're comparable; they're not
19 completely different.
20 But also, it's important to remember that the
21 production process here is likely to be a major part of
www.transcriptionetc.com
64
1 a company's IP around that product because the phages
2 you can't necessarily patent themselves if they're
3 natural. So one of the things you can have you can
4 patent, and also even protect the trade secret, is the
5 manufacturing process. And so you have to keep that in
6 mind when it's going through the regulatory process.
7 So the purification method that you use will
8 probably then determine that product identity. And
9 also, you have to remember that, because phages are
10 very diverse, not all phages are going to necessarily
11 respond the same way to the same manufacturing process.
12 So you may have to have, basically, a -- you might have
13 a set process that you're going to have to have
14 variances built in for or maybe even several different
15 parallel processes, depending on the phage you want to
16 produce. And they can vary in terms of their pH
17 stability and their surface charge, their stability in
18 various buffers, and formulation agents. And so you're
19 probably going to need to be able to have room for
20 adjustment depending on different phages from that --
21 from a library.
www.transcriptionetc.com
65
1 So just a note about phage production though.
2 So the technology really to maximize that yield would
3 really be a big help in this field. So this is a
4 picture from the Patterson case that we were involved
5 with, with the groups from UCSD -- and Dr. Schooley’s
6 here -- and the group from the Navy.
7 And so when phages were actually administered
8 to Dr. Patterson, they were administered as doses of 10
9 to the 9 pfu and diluted into 100 ml Ringers and
10 infused. So 10 to 9 pfu is actually not that much. We
11 can actually grow cultures of 10 to the 9 to 10 to the
12 10 pfu per ml in the lab. And you can grow up a couple
13 liters of that even just in the lab scale.
14 And so in theory, one liter of culture could
15 produce like a thousand doses of this phage for a
16 patient. But in practice, by the time we actually went
17 through our purification process, our recovery was
18 something like 0.1 percent. So we had to produce
19 multiple batches. So any technology that you can use
20 to -- if you can recover even half of the initial
21 particles that are in the -- that initial culture, that
www.transcriptionetc.com
66
1 makes it much more efficient than at least what we've
2 been able to do in the lab so far in a rush. So having
3 an optimized production process certainly helps.
4 So if we're looking at then producing phages
5 possibly of different strains, so the variation of the
6 host is also going to be a contributing factor to your
7 product identity. And it's going to determine partly
8 the contaminant profile. So if you had a very diverse
9 phage collection, you're likely going to have more than
10 one host you're going to have to grow all these guys
11 on. And so different strains of the same species, they
12 can vary in their toxin production, their LPS structure
13 and how pyrogenic it is. Mobile element content will
14 change. And also, many bacterial strains carry
15 multiple prophages.
16 And this is just a map from a relatively old
17 review. But this shows maps of -- these are actually
18 active prophages in bacterial genomes. And so it's not
19 that unusual for the average bacterial genome to carry
20 one to three functional prophages. And when you
21 propagate the phage, your phage, your virulent phage,
www.transcriptionetc.com
67
1 on that strain, some of that prophage will be induced
2 and will end up in that culture as well.
3 It likely is a very minority component. For
4 example, if you have 10 to the 10 of the phage you
5 want, you might have a 10 to the 4 or 10 to the 5 per
6 ml of this other phage. But it's going to be there,
7 and it's going to vary depending on the strains that
8 you want to use.
9 And these are an issue because they could
10 possibly, in theory, transduce DNA from your
11 propagation host into another host in the patient or
12 possibly between strains in the same -- in the patient
13 once they're administered. Maybe, right? These are
14 all just theoretical possibilities.
15 It is possible to take bacterial strains and
16 engineer them to reduce their contaminant profile. So
17 this is an example of the bacillus strains, the root of
18 strain 168 that as it's called MG1M, which is developed
19 by, I believe, a Japanese group. And really, they just
20 took the bacillus strain 168, which is kind of the go-
21 to model strain for bacillus subtilis, and they removed
www.transcriptionetc.com
68
1 all the prophages and the polyketide synthesis and a
2 whole bunch of other stuff they thought the cell didn't
3 really need. And they ended up removing about 23
4 percent of the genome.
5 And this was for a -- for using the cell as a
6 biofactory. They wanted to be able to put your own
7 metabolites in here. It will grow more efficiently and
8 make more of the enzyme or whatever it is that you
9 wanted. But you could theoretically do the similar
10 thing here for phage hosts as well and remove all the
11 components that you don't want from that genome. It's
12 not a -- it's a nontrivial undertaking to do this.
13 So if you had to do this, say, with 50 strains
14 to propagate your phage, that will be -- that would be
15 an undertaking but certainly doable. It also maybe is
16 possible to generate like a universal host or at least
17 a much broader host range host that you have. So
18 instead of having to take your hundred phages and
19 propagate them on 50 different hosts, you would
20 propagate them on five different hosts, which would
21 certainly simplify your manufacturing process. It's
www.transcriptionetc.com
69
1 also maybe possible that you could express phages from
2 a completely orthologous host. For example, in yeast,
3 they could be recombinantly produced in a yeast cell or
4 some other method. Again, I'm not aware of anybody
5 actually doing that, but maybe it's possible.
6 So product identity, also the -- in terms of
7 active ingredient concentrations, are important. So in
8 a phage product, the actual concentration of active
9 ingredients is likely to be very low, in the order of a
10 microgram per ml or less. So the kind of standard HPLC
11 peak or something you're looking at for a drug is
12 likely not going to work here. And so the way people
13 normally evaluate the potency of the product then is
14 turn to an active titer. You're looking for plaque-
15 forming units on a lawn of bacteria.
16 And so that's the way we have done it. As far
17 as I know, that’s the way other groups are doing it as
18 well. And so you actually are measuring infected
19 particles, which is nice. They have to be viable to
20 form a plaque. But you also have to be aware that the
21 number of plaques you'll get can vary a lot depending
www.transcriptionetc.com
70
1 on the host that you put the strain on and the plating
2 conditions.
3 So this is an example from our group of some
4 staph aureus phages. And you can see the plating
5 efficiencies can vary by half a log, depending on the
6 host you put them on. So you can get -- if you have
7 your phage product, if you titer on one host, they'll
8 tell you you have 10 to the 9. A different host might
9 tell you you have 5 times 10 to the 9.
10 And so you have to think about these plaque-
11 forming unit assays. It's kind of that's kind of the
12 minimum number of viable phages that are in there. If
13 you could find a better host that gives you more
14 plaques, then that would kind of -- that raises that
15 floor up.
16 Another issue then is product impurities, as
17 well. So crude lysate contains media components and
18 bacterial cell debris, and it's going to have endotoxin
19 in it if it's a gram-negative phage. And these range
20 from relatively harmless, right? You're going to have
21 some sugars in there. And also, it could be very
www.transcriptionetc.com
71
1 dangerous like endotoxin. And there has to be
2 something -- a decision arrived at really on what
3 impurities you can tolerate in the final product.
4 And this is probably going to vary depending
5 on how you're administering it. So if it's going to be
6 a phage product for oral or topical use, you could
7 probably tolerate more impurities than you would for
8 something which is going to be introduced parenterally.
9 And again, this is -- in, at least where I am in the
10 field, nobody really knows what the rules are right
11 now. So that's something that has to be developed, I
12 think in conjunction with the regulators and the people
13 who are actually manufacturing this to decide what are
14 the acceptable impurity profiles that are going to be
15 tolerable because that will then also drive the
16 manufacturing process as well.
17 The other issues around phages, which are not
18 really necessarily manufacturing-related, but they're
19 intellectual property related. So patent protection
20 for phages are difficult to determine. They are
21 products of nature. So engineered phages are certainly
www.transcriptionetc.com
72
1 more patentable than the natural ones.
2 There's a lot of prior art dating back to the
3 1920s. For example, there are patents that exist that
4 they have patented the idea of using phages to treat
5 bacterial infections, an idea which was invented in, I
6 believe 1917 or 1918. So it's more difficult, then, to
7 get intellectual property protection.
8 And if your product is actually a viable
9 phage, it's going to replicate in the host, you can't
10 necessarily rely on trade secrecy either because
11 anybody can just take your products and just culture
12 the phage out of it. And then, they've got the phage.
13 So trade secrets aren't going to work there either, so
14 some kind of intellectual property protection is still
15 -- that's still evolving.
16 They're very narrow spectrum. So I think it's
17 harder to get companies to get involved in this field
18 because you're developing very narrow-spectrum
19 treatments, which are -- you're not going to be able to
20 cross-market them very broadly, so that makes, I think,
21 a lot of companies reluctant to invest.
www.transcriptionetc.com
73
1 And really, the idea around this is the
2 traditional business model of a single mass-produced
3 antibiotic aren't going to work, of course, for
4 personalized medicine. But something we've been
5 hearing a lot lately in the last few years is about
6 just how the model of antibiotic development in general
7 -- a lot of companies have just gotten away from
8 developing antibiotics because they're not profitable,
9 because they're -- you're not going to have a large
10 market. They’re expensive to develop.
11 And if that's true for small molecules, it's
12 also true for this. And so if there's a fundamental
13 issue with the economics of this developing anti-
14 infectives, that's less of a scientific issue. I think
15 it's more of an economic or a policy issue. But it
16 also -- that's going to drive the availability of these
17 products.
18 So just to wrap up, so some solutions that --
19 to address some of these things, we need better
20 understanding of some of the non-paradigm phages that
21 are infecting different hosts. Like we know a lot
www.transcriptionetc.com
74
1 about a few phages that infect e coli and salmonella
2 but relatively little about most other phages -- and
3 also, really to be able to work in these ESKAPE
4 pathogens and more tools to be able to do molecular
5 biology in wild clinical strains, rather than just
6 model strains. There needs to be some more guidance, I
7 think -- and this is the purpose of this workshop, to
8 address issues around what are the production
9 standards, contamination profiles, how you define
10 product identity in this case, and how you develop IP
11 protection and that IP protection may actually help get
12 more investment in the field and, finally, just to
13 address that policy and economic landscape, really to
14 make actually any infective development just attractive
15 in general or to have it -- make it happen some way.
16 All right. So I would like to thank the
17 organizers and thank the audience. And I'll wrap up.
18
19 PANEL SESSION WITH Q&A
20
21 DR. TARAPOREWALA: All right. So we will -- I
www.transcriptionetc.com
75
1 now invite the speakers and the panelists to come join
2 us here in the front. We'll take a few questions from
3 the audience. And we may take a few questions online
4 if there are any. And please feel free to state your
5 name and affiliation. And also, keep your questions to
6 the point so that we stay focused on the topic of the
7 session, which is manufacturing.
8 Okay. So we have the speakers, and then we
9 have Dr. Roger Plaut from the Office of Vaccine
10 Research and Review and Ms. Anita Richardson from the
11 Office of Compliance and Biologic Quality, all in CBER.
12 And I don't see any questions and none online as yet.
13 So I will take the first stab.
14 So I was wondering if -- we talked about the
15 challenges. And I was wondering if maybe we could
16 discuss what are the advantages of -- what have we
17 learned from the rare and orphan disease drug product
18 development that can be directly applicable to
19 individualized therapeutics?
20 DR. GAO: So I believe the experience from the
21 ultra-rare disease or rare disease gene therapy can be
www.transcriptionetc.com
76
1 very helpful or informative to developing the
2 individualized therapy because I think, if you think
3 about the process of discovery or develop gene therapy
4 product -- it doesn't matter for ultra-rare, or rare,
5 or moderate-rare disease, or prevalent diseases -- I
6 think should be the same. And actually, the good thing
7 is, unlike large applications such as Duchenne’s
8 muscular dystrophy, each patient may need several, 10
9 to 15 or even up to 10 to 16 dose, vector dose, per
10 patient. But for those rare diseases, the population
11 is much bigger -- much smaller, and particularly ultra-
12 rare.
13 So the manufacturing burden is less. I think
14 current technology should be able to apply for
15 manufacturing. So that's advantage.
16 DR. TARAPOREWALA: Anybody else who would like
17 to take a stab?
18 DR. GILL: I have a question for Dr. Gao
19 actually. So for these ultra-rare diseases, is the
20 manufacturing mostly just lab scale at that point, or
21 are you still finding a CRO to try to make this stuff?
www.transcriptionetc.com
77
1 DR. GAO: Yes, this is a very interesting
2 question. Based on our experience as a gene therapy
3 center at UMASS and Rare Diseases Institute, so far, I
4 think for those kind of few patients, single patients,
5 or a couple patients, even academic viral vector core
6 should be able to address the needs. And we have been
7 very much enjoying this kind of collaboration with
8 academic GMP manufacturing core. I will say it's quite
9 -- much easier than bigger trials and bigger disease
10 population.
11 DR. TARAPOREWALA: We'll take one from the
12 audience.
13 GUY: So my name is Guy (phonetic). And I
14 don’t know if you can hear me. Is the microphone
15 working?
16 UNIDENTIFIED MALE: This one's working. You
17 can come back here if you want.
18 GUY: Oh, okay.
19 UNIDENTIFIED MALE: Just come back here. Use
20 this one.
21 GUY: Okay. It’s an individualized microphone
www.transcriptionetc.com
78
1 here. All right. So this is a question direct to
2 Professor Gao. You mentioned that manufacturing
3 capacity is clearly the challenge to produce AAV.
4 Obviously -- and if we can actually create a cell line
5 that can produce high titers of the AAV, you don't have
6 to do the transfection. So what exactly the challenges
7 are they and to produce such a cell line, if we
8 actually can do that and the manufacturing capacity
9 should be easily overcome? So I'd like to hear your
10 opinion on that.
11 DR. GAO: Thank you. I think you refer to AAV
12 production. So the major issue is that it's we need
13 regulatory protein and a capsid protein for
14 manufacturing process, particularly REP, regulatory
15 protein. That protein is very cytotoxic. And the
16 capability to generate stable integrated and producer
17 cell line, introduce the REP express under controllable
18 levels, particularly inducible levels, it has been a
19 major challenge. So that could be one issue.
20 The second issue is that other than
21 replication biology and packaging biology, AAV very
www.transcriptionetc.com
79
1 much depends on cellular factors to -- actually,
2 cellular factors contribute to all this replication and
3 packaging process. Our understanding of the cellular
4 factors that can enhance or improve gene therapy vector
5 production, we do not understand. I think if we can
6 understand the REP, where you establish your stable REP
7 cell line, as well as understand the cellular factors
8 and, importantly, if we can come up much tighter
9 inducible system that you can pharmacologically or some
10 other ways induce regulatory protein expression, that
11 will significantly give us opportunities to generate
12 so-called real producer cell line.
13 DR. TARAPOREWALA: Thanks. We can take the
14 next question from the audience.
15 MR. ALDRICH: Hello. Good morning. My name
16 is Steve or Stephen Aldrich. I'm 64 years old. I'm
17 the founder of a company called MyCancerDB. I am a
18 stage-4 adenocarcinoma of the esophagus patient.
19 I was diagnosed in March, late March of 2017.
20 For those of you who aren't familiar with that, most
21 people die for sure within two years. It's a very,
www.transcriptionetc.com
80
1 very deadly cancer.
2 Here I am three years later. Part of the
3 reason is that I was able, being a Harvard-trained
4 biologist, to recognize that if there was an answer to
5 my problem , because the prognosis is -- or the
6 prescriptive care is strictly palliative, that it was
7 going to be found in my data. And so I had -- was very
8 fortunate to be able to have all of my fundamental
9 sequencing data, my cancer genome, my healthy genome,
10 my microbiome, et cetera, and the related datasets,
11 done. And with that information, I was able to
12 leverage it to, first, identify a clinical trial that
13 kept me alive for a while.
14 And while that was going on, I had designed,
15 tested, and manufactured a fully personalized
16 neoantigen peptide cancer vaccine. And as a
17 consequence of going through that, I realized that the
18 current system was broken and that we needed to be able
19 to accelerate access to this kind of fully personalized
20 N of 1 therapy to the 1,700,000 patients who were
21 diagnosed with cancer in the U.S. last year and
www.transcriptionetc.com
81
1 especially the 610,000 that died.
2 So a couple of points I want to make about the
3 panel. The first is that, in order to enable the kind
4 of gene therapy that was presented, we have to
5 completely reengineer the entire system. And that
6 reengineering starts with the fundamental genetic data
7 of the patient, the sequencing data.
8 Clearly, to all your presentations, it's the
9 sequencing data that's fundamental. Whether it's for
10 phages or for people, it's the sequencing data that we
11 must have. But who controls that data?
12 If that data is the most important
13 personalized information about an individual, shouldn't
14 we create an infrastructure that keeps control and
15 ownership of that data with the individual? I just
16 thought I'd bring that to the panel and ask them for
17 comment. I think what we're doing at MyCancerDB is
18 putting in place that infrastructure.
19 And the idea is to be able to leverage that
20 information to virtualize the supply chain in a way
21 that shrinks the cost, speeds up the development
www.transcriptionetc.com
82
1 process, and maybe does something about the 610,000
2 people who died of cancer. So I -- last year, which is
3 12 times the number that died in the Vietnam War. I'd
4 be interested in your comment.
5 DR. TARAPOREWALA: So I guess that was a
6 comment, maybe not a question. And if nobody in the
7 panel has a comment towards that, we might move on to
8 the next question, if there's one from the audience.
9 MR. CAMPBELL: Hi. This is Joe Campbell from
10 NIAD. I have a question for Dr. Gao. I was wondering
11 -- you mentioned the advantage of flipping the ratio
12 from 90 to 10 to 10 to 90. And that obviously would be
13 a great advantage. And I'm just wondering if you have
14 any thoughts on whether -- what's likely to be
15 important, the vector, the payload, or the strain in
16 which you grow it in? And I guess one further
17 question. Have you ever, thought about making the
18 unfilled vector have something like an endonuclease, a
19 very rare endonuclease cut site that the filled one
20 wouldn't have, so that you could select against it?
21 DR. GAO: Very interesting questions. So this
www.transcriptionetc.com
83
1 flip from 90 to 10 to 10 to 90, it's quite challenging.
2 Actually, if you ask me the answer, I don't have the
3 answer. But I know where to go.
4 That is what I just talked to the first person
5 who asked the question from [inaudible]. It's the same
6 issue. It’s we have to understand how AAV replicates
7 and packages.
8 And we have to understand how cellular factors
9 can help us. Because AAV itself, it's almost a virus.
10 It needs a lot of things from host and from the helper
11 virus. So that is definitely one thing to consider.
12 Second thing is the purification methods. If
13 we are defective -- I mean virus is defective to
14 packaging for particles. But if we have a scalable
15 GMP-compatible process to separate empty from full,
16 then we should be able to enhance the potency of our
17 drug, reduce immunogenicity of those useless empty
18 particles.
19 DR. TARAPOREWALA: Okay. Thank you. We'll
20 take the next question from the audience.
21 MR. KELLY: Good morning. My name is Matt
www.transcriptionetc.com
84
1 Kelly from Sarepta Therapeutics. I just had a follow-
2 up question on the empty/full ratio.
3 DR. TARAPOREWALA: Could you speak up a little
4 bit because we cannot hear you.
5 MR. KELLY: I just have a follow-up question
6 on the empty/full ratio. As Dr. Gao highlighted, a 90
7 percent full count would be acceptable. From an FDA
8 perspective, a commercial scale, is this also an
9 acceptable reality for an approval?
10 DR. TARAPOREWALA: So I think at this point,
11 we are taking questions to see what the researchers
12 think in how to develop individualized therapeutics.
13 And I think Dr. Gao just mentioned that the empty
14 ratios does affect the potency and does affect the
15 purity and the immunogenic load. So I think that
16 applies rather product is in clinical trial or being
17 developed for licensure. So I think his answer is
18 pretty broad and comprehensive in that respect.
19 MR. KELLY: Okay. Thank you.
20 DR. TARAPOREWALA: We'll take the next
21 question from the audience.
www.transcriptionetc.com
85
1 MS. HESTERLEE: Hi. This is Sharon Hesterlee
2 from the Muscular Dystrophy Association. So I
3 appreciate very much all the improvements in process
4 development that have helped increase the yield in
5 manufacturing. But I'm wondering if we're going to hit
6 a limit at some point where we really can't further
7 increase the yield in a practical way. And should we
8 be looking at things more like targeting? You
9 mentioned early on that that was one of the issues.
10 For example, AAV9, if you're trying to target
11 the brain, the vast majority, if you go systemic, is
12 still going to liver. So I could see that you might
13 get an order of magnitude improvement if you could just
14 better-target vectors rather than increase
15 manufacturing. Is that something to consider?
16 DR. GAO: Yeah. That is a great question.
17 Actually, that's the whole field of gene therapies and
18 particularly AAV gene therapies. We have been
19 struggling for 50 years. That's what we're trying to
20 do.
21 But what I have to tell you is I think recent
www.transcriptionetc.com
86
1 research demonstrated what is achievable. I refer to
2 Dr. Ban Deverman's publication in Nature Biotech a
3 couple years ago. I think he demonstrated that you can
4 accomplish that more efficient cell type, tissue type,
5 targeting. But the only issue we're running into now,
6 soon afterwards, we realized what you select against
7 will be what you get out of the selection system.
8 So previously, the selection was done in
9 mouse. And we generated the super mouse CNS vectors.
10 So the next stage, I think as was discussed by Dr.
11 Marks, that we look into potential humaned mice, as
12 well as probably primates, animal models, to do that
13 selection. I think that people know how to do it, just
14 have to switch the system.
15 MS. HESTERLEE: Thank you.
16 DR. TARAPOREWALA: We'll take the next
17 question from the audience again.
18 MR. MCFARLANE: Thanks. Richard McFarland for
19 the Advanced Regenerative Manufacturing Institute. So
20 as Dr. Gao said for the hyper-rare diseases, you think
21 maybe laboratory scale manufacturing would be
www.transcriptionetc.com
87
1 sufficient. It seems that raises -- we've done a lot
2 of talking about traditional CMC, but it seems that
3 also raises questions about distributed inspection and
4 regulatory compliance for those centers. And I wonder
5 if you can share any thinking about what you've had
6 about how you're going to inspect those facilities if
7 they do exist across multiple hospitals.
8 DR. TARAPOREWALA: You can go ahead, and then
9 Anita can weigh upon it. Go ahead, Dr. Plaut.
10 MS. RICHARDSON: Richard, thank you for that
11 question. And I'm not sure I have a simple answer. I
12 think that we are looking at those types of facilities
13 on a case by case basis, taking all the facts and
14 circumstances and the product into consideration and,
15 also, the flexibility needed in this field.
16 MR. MCFARLANE: Well, Anita, I'm glad you
17 don't have a simple answer because I don't think it's a
18 simple solution. But I think it's something that I
19 know we're thinking about and I think others may be
20 thinking about. So maybe it's a place for focus
21 workshops or guidance, something in the future to help
www.transcriptionetc.com
88
1 think through those issues as you go from an IND to a
2 true manufacturing facility.
3 DR. TARAPOREWALA: Okay. We'll take the next
4 question from the audience again.
5 MS. GALEMBO: Marian Galembo from BiomX
6 (phonetic). And I will talk about phage therapy. So
7 Dr. Gill has really put forward a whole lot of the
8 dilemmas and the problems that we may be facing with
9 individualized therapy and manufacturing in hosts that
10 could be derived from individual patients and each one
11 with different characteristics. I would like to hear a
12 bit about the Agency's approach on how to handle those
13 differences and variations in terms of impurity
14 characterization, if there is any thoughts about that.
15 DR. PLAUT: So the goals of this workshop are
16 to get feedback from the community, from stakeholders.
17 So we are not really prepared to make any public
18 statements about our -- how we're regulating these
19 products or how we intend to regulate these products.
20 I can tell you that every decision we make is based on
21 science, based on evidence. And we ask sponsors or
www.transcriptionetc.com
89
1 potential sponsors to come to us and provide all their
2 data and to propose to us what they think their
3 specifications should be, what they think the limits
4 should be.
5 And then we have a discussion with the
6 sponsor. This is one reason why we think it's very
7 important that sponsors interact with us as early in
8 the process as they can. We're happy to have these
9 sorts of discussions in -- early on in development.
10 But again, I think it's important for you to understand
11 that we have an open mind, and we make our decisions
12 based on science and evidence.
13 MS. GALEMBO: Thank you.
14 DR. TARAPOREWALA: Thank you. We'll take the
15 next question from the audience again.
16 MS. WALKER: Hi. My name is Karen Walker.
17 I'm from Genentech. I want to thank you both for such
18 good talks.
19 I actually have a question for both of you.
20 In each of your cases, you talked about manufacturing
21 challenges from an intellectual property perspective
www.transcriptionetc.com
90
1 and a capacity perspective. And Dr. Gao, you talked
2 about the large role that CGMOs are playing in the
3 field.
4 And given that these are individualized
5 therapies and given that manufacturing platforms and
6 establishment of these platforms would drive down cost,
7 what are your ideas about how we could open up access
8 to those platforms so that we could really drive
9 targeted individualized therapy treatments to patients
10 with predictable outcomes?
11 DR. GAO: So I have to say, not myself, but
12 two weeks ago, NCATS and P.J. Brooks, and other
13 colleagues from NIH, they held a meeting on AAV gene
14 therapy manufacturing. And that meeting, their whole
15 purpose was really understand the landscape, understand
16 the challenging, understand to kind of brainstorm this,
17 all the scientists, and regulator and administrator to
18 figure a way to ask -- to address that question. So I
19 think I don't know whether P.J. is still here or not.
20 He was here in the morning.
21 They do have actually, in after reading
www.transcriptionetc.com
91
1 communications, I think they gradually -- is in the
2 process of formulating some kind of solutions for
3 ultra-rare, rare, giving these individualized
4 therapies.
5 DR. TARAPOREWALA: Dr. Brooks, would you like
6 to comment on that?
7 DR. BROOKS: Yes. Thank you for that. Thank
8 you, Guangping. At NCATS, we are certainly interested
9 in these kind of broad issues and making these
10 platforms more widely available for these types of
11 treatments. And, specifically, our -- the Cures
12 Acceleration Network, which is part of the NCATS
13 Council, is very much focused on addressing this issue.
14 And part of the goal of the meeting that we
15 had a couple weeks ago was to identify the
16 opportunities that might be available to make certain
17 investments in research approaches to really bring this
18 forward. So it is something that is under active
19 investigation, by NCATS as well as with other -- our
20 other partners within the NIH and industry, and of
21 course the FDA as well.
www.transcriptionetc.com
92
1 MS. WALKER: Thank you. And just maybe a
2 comment. I think this also speaks to the fourth bullet
3 point on the slide, which is it would be interesting
4 for FDA to think about additional guidance on how we
5 could have regulatory flexibility around establishing
6 production platforms for these individualized
7 therapies, which then would change the focus of the
8 review more to the therapy itself and the safety and
9 efficacy of those, where we could maybe control some of
10 the variability around the platforms.
11 DR. BROOKS: That issue might come up later on
12 in the afternoon as well.
13 MR. ALDRICH: Can I just follow up that
14 excellent question and the point? There was actually a
15 question in my earlier comment. And it has to do with
16 -- for the reengineering that we're going to have to do
17 around manufacturing, there are methodologies that are
18 well-known to the FDA that are used in the regulation
19 of, for example, the food industry, where you use
20 hazard and critical control-point methodologies to
21 regulate, in a scientific way, the safety of a
www.transcriptionetc.com
93
1 particular supply chain.
2 And I -- that is something like that, where we
3 substitute the regulation of the end product as a
4 standard product, with the regulation of the process
5 that we use to get to that end product, is inevitably
6 the endpoint, in my opinion, of enabling N of 1
7 therapies to come to fruition. And this isn't a
8 hypothetical issue for me. This is a very concrete,
9 real issue for cancer patients everywhere, when it
10 comes to something like access to fully personalized
11 neoantigen peptide cancer vaccines.
12 So I'd just like you to comment on -- perhaps
13 our two representatives from CBER -- could comment on,
14 is there an awareness within the FDA of this idea of
15 using well-known GMP and HASIP-related methodologies to
16 -- as a basis for approving an “N of one” manufacturing
17 process, and if not, why not?
18 DR. PLAUT: So again, I will say that we are
19 openminded about these different approaches to these
20 kinds of therapies. And we see that the IND process,
21 as it stands now, is appropriate for the process for
www.transcriptionetc.com
94
1 getting products into the clinic. For as far as how a
2 product or a process could be -- could lead to
3 licensure, that is something that will remain to be
4 seen. We don't really have any sort of policy
5 statement to make at this point. But again, we have an
6 open mind about these issues.
7 DR. GILL: I'd just like to make a point. So
8 on the phage side, that's sort of -- I think what we're
9 looking at, is that you have to have kind of the
10 process approved rather than individual phages.
11 Because at least the way the field is moving right now,
12 if you're looking at, say, a personalized therapeutic,
13 say, for Acinetobacter baumannii, there's going to be a
14 collection of 200 phages or 500 phages. You're not
15 going to have detailed preclinical and clinical data on
16 all of those phages. You're going to have maybe on the
17 subset of those.
18 And then the rest of them are going to be kind
19 of carried along, based on some criteria that are set -
20 - that they look like the phages that work and so on.
21 Because I think if you require that every phage goes
www.transcriptionetc.com
95
1 through a full clinical evaluation, then it's really
2 never -- I don't think it's ever going to happen
3 because you can't run clinical trials on 200 individual
4 products.
5 DR. TARAPOREWALA: In the interest of time,
6 because we have a very good number of sessions, I urge
7 you to -- the two attendees that have gotten up to ask
8 questions, I'd urge you to get hold of the speakers
9 offline -- they will be around here for the whole day -
10 - and perhaps ask the question. So in the interest of
11 time, let's thank the speakers and the panelists for a
12 very nice session. Thank you.
13 DR. RAYCHAUDHURI: So I'd like to thank all
14 the speakers, the panelists, and the audience, for a
15 stimulating discussion. And I'm really sorry we had to
16 cut it short. But please do take advantage of the
17 speakers being here to speak with them one on one.
18 We'd like to now take a short, maybe 10-minute
19 break, just to stretch your legs, and we will reconvene
20 with session 2. Thank you.
www.transcriptionetc.com
96
1 [BREAK]
2
3 SESSION 2: TOOLS FOR SAFETY TESTING AND DEVELOPMENT
4
5 DR. RAYCHAUDHURI: Okay. Could everybody
6 please take your seats because we'd like to get Session
7 2 started? So that was a short break. And I hope you
8 had a chance to stretch your legs. We are ready to
9 start Session 2.
10 I'd like to introduce Dr. Sandhya Sanduja, who
11 will moderate the session. Dr. Sanduja is acting team
12 lead for Pharmacology Toxicology Branch I in the
13 Division of Clinical Evaluation, Pharmacology, and
14 Toxicology in Office of Tissues and Advanced Therapies.
15 Dr. Sanduja will moderate this session, which is on
16 nonclinical evaluation and tools for safety testing for
17 individualized therapeutics.
18
19 SESSION 2 MODERATOR INTRODUCTION: DR. SANDHYA SANDUJA
20 (CBER)
21
www.transcriptionetc.com
97
1 DR. SANDUJA: Thank you, Gopa, and thank you,
2 everybody who's present here as part of our workshop to
3 facilitate end-to-end development of these
4 individualized therapies. So after a very interesting
5 and engaging Session 1 on manufacturing of these
6 individualized therapies, it is my pleasure to take you
7 to our Session 2, which will focus on tools for
8 preclinical testing of these individualized therapy
9 products that are regulated by CBER. The session will
10 begin with presentations by our distinguished speakers,
11 Dr. Albert Seymour from Homology Medicines, Dr. Malachi
12 Griffith from Washington University School of Medicine,
13 and Dr. Keith Joung from Harvard Medical School. These
14 presentations will be followed by a panel discussion,
15 which will be led by our speakers and our CBER subject
16 matter expert, Dr. Zuben Sauna, and our audience.
17 All right. So to give you a high-level
18 overview for this session, let me begin with this
19 schematic, which shows the traditional drug development
20 pathway. The preclinical testing paradigm for
21 individual therapeutic products looks very similar to
www.transcriptionetc.com
98
1 these other therapeutic modalities, where efforts are
2 spent during the discovery phase. They focus on
3 identification and optimization of the lead candidate,
4 followed by IND-enabling preclinical studies that
5 evaluate and establish proof of concept and safety of
6 these products.
7 However, when we talk about individualized
8 therapeutic products, the key distinction -- obvious
9 distinction that these products have are that they are
10 patient-specific. They are patient-tailored products,
11 and testing a healthy individual is either not feasible
12 or not ethical. So this is where the standard paradigm
13 of drug development won't apply.
14 During our session, we will hear from each of
15 our speakers some of the key challenges that are
16 encountered during preclinical safety testing and
17 development of these individualized therapeutic
18 products. We'll also hear from them what are some of
19 the opportunities we have to address these challenges,
20 including some of the regulatory -- the existing
21 regulatory framework to address these challenges, and
www.transcriptionetc.com
99
1 these challenges and opportunities, how they are
2 perceived by the developers as well as regulators. So
3 as we know, these products are patient-specific. These
4 are tailored for either one or a small number of
5 patients, instead of being tested in healthy
6 individuals.
7 So this brings in an inherent challenge
8 associated with these therapeutic products. And what
9 we expect from a preclinical testing program for these
10 individualized therapy is, that since these are going
11 into one or a small set of patients, to strike a
12 balance or to strike that favorable benefit-risk
13 profile, preclinical evidence to support the rationale
14 for safety of administration of these products and
15 patients. Such support must come from one or more
16 preclinical studies that are conducted in a relevant
17 animal model of disease using the intended route of
18 administration, since for these individualized
19 products, many times, these can be rare or ultra-rare
20 diseases, where the only patient population is
21 pediatric and in those cases, preclinical studies need
www.transcriptionetc.com
100
1 to establish prospect of direct benefit, as is required
2 by Code of Federal Regulation.
3 So fulfillment of all these parameters can get
4 extremely challenging in the absence of a relevant
5 animal model of disease. When there is actually a
6 relevant animal model of disease, there may be other
7 challenges associated with preclinical development of
8 these products.
9 And these challenges may vary. They may arise
10 on a case by case basis and may include appropriate
11 dose-level extrapolation, delivery with respect to the
12 intended clinical route of administration, the
13 procedure and device, as well as informing the
14 appropriate clinical monitoring for the clinical trial.
15 For in silico or computational tools that are used for
16 preclinical testing, there are challenges that lie in
17 validation of these algorithms to test -- and other
18 test methods for robustness, whether they can
19 confidently and reproducibly perform to the desired
20 standards of safety and activity of these personalized
21 therapeutic products.
www.transcriptionetc.com
101
1 Some additional challenges, which can arise
2 later in development, may include additional -- conduct
3 a requirement of additional nonclinical studies that
4 may be needed, depending on if there are significant
5 manufacturing changes that happen during later part of
6 development, also for the potential for development and
7 reproductive toxicities that may be present. So these
8 challenges, which may actually seem as hurdles to begin
9 with, may not necessarily be roadblocks to development
10 of these individualized therapeutic products. Instead,
11 they may be taken as opportunities which can drive
12 innovation, so particularly, with these products, where
13 we know there is so much science-based evidence in
14 development.
15 So they can drive innovation, particularly
16 advancement, in science and technology for better
17 models and approaches to preclinical testing. For
18 example, in silico methods that are used to inform
19 safety and activity of products can have adaptive
20 design to them. And as our experience -- clinical
21 experience grows with these products, that adaptive
www.transcriptionetc.com
102
1 design can be -- is incorporated in the tools that are
2 used to inform safety and activity.
3 Similarly, innovation happens when novel in
4 vitro methods are being developed. There can be -- we
5 have seen examples where patient-derived IPSCs or
6 patient-derived organoids can be used, particularly
7 when a relevant animal disease model is not available.
8 And there, such in vitro models can actually allow for
9 reduction, refinement, and replacement of animal
10 testing. Innovation can also play a role in developing
11 relevant and more robust in vivo models for testing
12 safety and activity of these products.
13 Next is collaboration. These hurdles can
14 actually be instrumental in driving collaborative
15 efforts, knowledge sharing in the community. So
16 instead of reinventing the wheel, as Dr. Marks alluded
17 to earlier, every single time, especially when using a
18 similar vector or similar platform technologies,
19 stakeholders may have the opportunity to leverage
20 existing preclinical and clinical data. That way, we
21 are harmonizing our efforts and leveraging data that's
www.transcriptionetc.com
103
1 already available to facilitate a faster translation to
2 the bedside.
3 And finally, these hurdles are instrumental
4 again in driving engagement with regulators, in
5 basically interactions with regulators to discuss the
6 preclinical program and how we can all agree and come
7 to terms for a feasible path forward for a specific
8 product. So with our session -- as we start the
9 session, after hearing from our speakers, I think we
10 all will be engaging in a discussion where the key
11 points to discuss would be what are some of the
12 challenges that are associated with preclinical testing
13 and development of these products including: platform-
14 based versus any product-specific concerns, use of
15 computational tools in safety testing of these
16 products, how leveraging existing nonclinical data
17 across similar products can be done, and regulatory
18 approaches and flexibility to preclinical testing of
19 these products.
20 So without any further delay, I would like to
21 welcome our first speaker, Dr. Albert Seymour. He's a
www.transcriptionetc.com
104
1 chief scientific officer of Homology Medicine, a
2 company that's developing gene therapy and gene editing
3 technology to treat patients for rare diseases. He has
4 a Biology from University of Delaware, an MS in
5 Molecular Biology from Johns Hopkins University.
6 And he received his PhD in post-doc training
7 in human genetics at the University of Pittsburg. He
8 has more than 20 years of experience in taking human
9 genetics to pharmaceutical RND, resulting in delivery
10 of multiple therapeutic programs into development. So
11 we welcome you and looking forward to your
12 presentation.
13
14 PRECLINICAL APPROACHES/CHALLENGES IN DEVELOPMENT OF
15 INDIVIDUALIZED THERAPEUTICS - DR. ALBERT B. SEYMOUR
16
17 DR. SEYMOUR: Thank you. Thanks, everybody,
18 and first, thank the FDA for giving me this opportunity
19 to come and talk a little bit about some of the
20 preclinical approaches and challenges to developing
21 these individualized therapeutics. I have to admit,
www.transcriptionetc.com
105
1 when I got the invitation, I really sat and gave it a
2 lot of thought, working with my colleagues as well, for
3 what are some of these challenges? Because we've been
4 focusing a lot on bringing therapies to rare, ultra-
5 rare disorders. And so I think a lot of those unique
6 opportunities and challenges fall also within very
7 individualized therapies.
8 And so I sort of pulled a collection of these
9 together. So one of the things I looked at right away
10 is some of the work that actually came out of Dr. Yu’s
11 lab out of Children. And this was -- really got a lot
12 of press. But it really exemplified that, where
13 there's a need, there is an opportunity to bring these
14 forward. And this was when publishing in New England
15 Journal of Medicine with this patient-customized
16 oligonucleotide therapy for a young child with a form
17 of Batten's disease.
18 And in ten months, going from sequencing,
19 identifying the unique mutation within that patient to
20 having an oligo that could actually correct that to
21 them being able to treat that patient, was just
www.transcriptionetc.com
106
1 fascinating to actually read and see how they were able
2 to go through that in leveraging some of the aspects
3 around cell-based models from that individual patient
4 because there wasn't an in vivo model. But then you
5 could also pull some information that was also
6 available on just the chemistries of the unique
7 oligonucleotides.
8 So with that, there are a lot of platform data
9 that can be leveraged to improve the efficiency. But
10 as we were thinking about this, there are really three
11 buckets that go into this, that the fundamentals of
12 benefit and risk assessment must still apply. We are
13 still bringing therapies forward to treat human
14 patients.
15 The other aspect is -- you heard from some of
16 the earlier speakers, so Dr. Gao, and then some of the
17 work that we're doing at Homology, there’s single
18 administration versus chronic dosing. So something
19 like AAV, it's a single administration. It brings, I
20 think unique challenges to an individualized patient
21 because of the ability you really can only dose once.
www.transcriptionetc.com
107
1 You really have to get it right that first time versus
2 something that you can do with chronic dosing. Perhaps
3 you can come in with a much lower dose and work your
4 way up to look for activity.
5 And then finally, with that, there are very
6 modality-specific requirements. And I'll walk through
7 some of these and exemplify these as I go along. So as
8 I think about these considerations, establishing the
9 fundamentals of benefit and risk, I really broke it
10 down into two categories, product-specific -- and what
11 I mean by that is that the actual -- if you're going to
12 replace the gene, the actual molecule that you're going
13 to replace versus a platform. So this would be, say,
14 if you're looking at a capsid AAV9, AAV15, AAV2 -- that
15 could be, can you leverage some of that that's more
16 platform-driven.
17 So from a product-specific perspective, you
18 still have to establish some form of biological
19 plausibility. This could be proof of concept in a
20 model, if one exists. But a lot of times, the timeline
21 that you need is not there to develop a new model. So
www.transcriptionetc.com
108
1 you may have to go to more in vitro models.
2 You also have to think about projecting the
3 dose for the clinical testing. Can you establish a
4 minimal efficacious dose? I think a really important
5 aspect is to identify dose response relationship early
6 and establish that pharmacology, particularly as you're
7 moving into that patient because a lot of these
8 patients may not have characterized natural history.
9 So you're going to want to know, if your drug gets into
10 that patient, are you actually engaging with the target
11 and seeing a pharmacology that you would be expecting.
12 And then finally, the safety margin would be very
13 product specific.
14 From a platform perspective, establishing the
15 delivery vehicle biodistribution. So there are a lot
16 of studies out there that perhaps you could cross-
17 reference if existing biodistribution are using AAV
18 capsid, for example. If you have biodistribution of
19 the capsid itself, say, in non-human primates or other
20 relevant species, can you leverage that when you come
21 in with another product but using the same AAV-based
www.transcriptionetc.com
109
1 capsid delivery?
2 And then there are also other aspects around
3 class-specific effects. So with the oligonucleotides,
4 the example that I just brought up from Dr. Yu at the
5 Boston Children's with the Batten's patient, really
6 looking at the underlying backbone of the chemistry.
7 Can you leverage what was already out there to then
8 bring that forward to help eliminate or reduce some of
9 the risk that may be associated with moving that
10 forward?
11 So upon establishing biological plausibility,
12 it's really characterizing the impact on this. And
13 this, I think, is really unique to the specific product
14 on that biological plausibility. What I mean by that,
15 if we're going after a monogenic disorder, we know what
16 the mutation is. We may be able to characterize that.
17 Are there cell lines that harbor that patient-specific
18 mutation? And ideally, depending on the therapy, is
19 there an animal model of that particular disease?
20 Because there's several things you'd want to
21 test: will the therapy reverse the effect of the
www.transcriptionetc.com
110
1 mutation? Will we see an increase in a protein
2 activity for something where that activity is missing?
3 Or can you decrease that activity in something that has
4 what's called a gain of function type of mutation?
5 So utilizing that to characterize the activity
6 establishes the dose response, understand durability of
7 that. And then I think one thing I really want to
8 bring up is establishing these endpoints that perhaps
9 you can test preclinically, that are the same aspects
10 and the same endpoints that you would test clinically
11 as you move into that, even into an individualized
12 patient. I want to use here just one -- an example of
13 that and establish a biological plausibility.
14 Here's the gene therapy for a disease called
15 PKU. There fortunately is a mouse model for this
16 disease. The mouse is missing or has a mutation in the
17 same enzyme that causes a human disease, phenylalanine
18 hydroxylase.
19 In this mouse, when you have two missing
20 copies of that or two mutations on that, you -- that
21 mouse cannot metabolize phenylalanine, very similar to
www.transcriptionetc.com
111
1 the human condition. So we can use that mouse to
2 actually do a single administration, and then see not
3 only do we reduce phenylalanine, but we can see that
4 loss durably over the lifetime of the mouse. So this
5 really helps establish the biological plausibility of
6 that specific product. We're adding back in
7 phenylalanine hydroxylase. We see it in vivo in an
8 animal model.
9 Here's another example, trying to understand
10 the dose relationship. So this is in a different
11 disease. This is a disease called metachromatic
12 leukodystrophy. It's a rare genetic disorder of
13 lysosomal storage.
14 These patients are missing an enzyme called
15 arylsulfatase. And unfortunately, when they're missing
16 that enzyme, that enzyme typically works in a lysosome
17 that is responsible for breaking down a product called
18 sulfatides. What happens in these children is these
19 sulfatides increase, and that increase in sulfatide is
20 toxic to the cells.
21 We can reference human data from people that
www.transcriptionetc.com
112
1 are out there that have roughly 10 to 15 percent of
2 normal human ARSA activity. There's a common
3 polymorphism within the ARSA that results in that.
4 These people generally are normal. And so we know that
5 at least if we can hit a 10 to 15 percent of a normal
6 activity, that should get us into something that would
7 be therapeutically relevant. And what you see on the
8 right is looking in a knockout model of that, showing
9 that as we increase doses with those, we can start to
10 see an increase in that activity in the brain trying to
11 shoot for that target threshold of 10 to 15 percent,
12 based on human data.
13 The other aspect is really trying to project a
14 starting dose for the clinical administration. So
15 again, this is unique to the specific product, but
16 there might be some platform-specific challenges and
17 opportunities to overcome that. So using AAV
18 specifically, AAV really does require in vivo data to
19 project a dose. You see differences in how AAV works
20 from an in vitro to an in vivo setting.
21 So here, you can either use an animal model of
www.transcriptionetc.com
113
1 human disease, if one exists. That will allow you to
2 assess the effect on the actual underlying mutation
3 that's similar to the example that I just gave on
4 introducing phenylalanine hydroxylase into ENU 2 mouse
5 model of PKU. So you can assess either an increase in
6 activity there or a decrease in vivo.
7 However, if you don't have an animal model, at
8 least of the things you can do is look in referencing
9 different data. So if you're using a specific delivery
10 approach such as AAV, if we already have that data from
11 other studies, can we reference that? Because that
12 will give us a sense of different doses, how many
13 vector genomes, at least we get from the capsid,
14 delivery into the liver, into crossing the blood/brain
15 barrier into the CNS. Can we leverage that to move
16 those forward?
17 So along these lines, again, unique to the
18 specific product, platforms that require repeat dosing,
19 you can utilize these models either in vitro or in vivo
20 to establish a dose response, to define that minimal
21 efficacious dose, and perhaps to develop biomarkers
www.transcriptionetc.com
114
1 indicative of pharmacodynamic activity, meaning that
2 your drug is engaging with the target and then having
3 the effect pharmacologically that you were expecting
4 coming in. For single administration platforms,
5 meaning a single dose, you can do -- all the first
6 three are the same. The only thing is for that fourth
7 that's very unique for these is that establishing the
8 starting dose that has a high likelihood of benefit.
9 Because particularly, with AAV or other types
10 of these approaches, the ability to re-dose at this
11 point is challenging. And so if you're coming into an
12 individualized person, you really want to make sure
13 that that first dose is going to have the activity that
14 you want. So that sort of raises the bar a little bit
15 form an AAV perspective. But the opportunities are
16 there, I think, to utilize different models to begin to
17 address that.
18 Just to exemplify that, again, I show a lot of
19 PKU data because we have a program that just is in
20 early clinical programs right now for that. Here's
21 some of the data from a 28-day dose ranging finding
www.transcriptionetc.com
115
1 study in ENU 2 mice. And so what we're able to show is
2 characterizing both males and female mice, looking
3 across a variety of doses, looking at a reduction in
4 phenylalanine. So that's a direct measure of
5 increasing PAH activity. So we're putting back in
6 something that was missing.
7 We can pick a dose that identifies that. And
8 then we can also look at another marker, which is
9 called tyrosine. So tyrosine is the direct product of
10 phenylalanine metabolism via phenylalanine hydroxylase.
11 And so we have two approaches that we can look at to
12 help identify what are those doses, at least in the
13 mouse model, that may deliver the effect that we're
14 looking for.
15 And then taking that information, we could
16 actually apply some PKPD modeling. And so we can
17 utilize those models to try to get an assessment of, if
18 we go into a patient population that, say, has a
19 baseline serum phenylalanine levels of 1,200, 1,800, or
20 even all the way up to 2,400, are there doses that,
21 based on the preclinical data, would deliver that kind
www.transcriptionetc.com
116
1 of result. We can then take the modeling and identify
2 and make some predictions around what percentage or
3 proportion of subjects would we expect to have that.
4 So based on this modeling, it's based on mouse
5 efficacy data. So what's unique about these mice, they
6 are homozygous for a genotype phenotype. We know that
7 if we go into the general population of humans, they're
8 not usually homozygous for a specific genotype, so it's
9 much more heterogenous.
10 But it really can allow us to select a dose
11 range that we know would have activity. And then we
12 can align that up with what we're seeing from our GLP
13 toxicology studies to assess where is our safety
14 margin, if you will, to make sure that these can fall
15 into that. I think the translation to humans is
16 required to fully characterize these, just given the
17 inherent variability of what you would see in human
18 populations compared to what you see in a very well-
19 controlled mouse and animal model environment.
20 So then moving along is really establishing
21 that safety margin. Here's where some of the, I think,
www.transcriptionetc.com
117
1 challenges come that would be unique to the specific
2 product. I know that if you're using one particular
3 AAV, the inside of that AAV is different as you go
4 across different diseases. And we know that any kind
5 of safety effect can be both based on the delivery
6 vehicle but also based on the actual API, if you will,
7 that's going to be delivered to that.
8 So you can use a model of human disease of
9 what's the safety margin, aligned with the
10 pharmacology. That's where, if you do have a model, it
11 really gives you a lot of that information, seeing
12 assessment or pharmacology, but then you can also see
13 assessment of where that safety margin is to assess
14 that. If you don't have a model, you can still use
15 wild-type animals to get a sense of the safety margin,
16 as well as the biodistribution of the therapy.
17 And then finally here, you can pull in some
18 reference studies particularly around the delivery
19 vehicle, so whether it's an AV capsid where you can
20 utilize that or some of the known class effects. What
21 I mean by here is -- again, I will use AAV as an
www.transcriptionetc.com
118
1 example. We know that AAV, there's immune responses to
2 it, and a lot of those are cytotoxic t-cell mediated,
3 against the liver. And so one of the main tox organs
4 is the liver. So you can reference that from other
5 studies to then guide you as to what to monitor, at
6 least as you think about moving into the clinic.
7 As far as safety and biodistribution, again,
8 from a product-specific, there could be mechanism-
9 driven safety. So what are the expression levels of
10 the specific product? What are the level of inhibition
11 that you need? If you inhibit 50 percent, you might be
12 safe. But perhaps, if you inhibit too much, you might
13 have some safety consequences and vice-versa on the
14 cell-specific challenges around maybe expressing too
15 much.
16 I think Dr. Gao brought that up a little bit,
17 as one of the last hurdles in getting into the clinic
18 is trying to regulate this. Some cells, based on
19 delivery, you may hit them very hard, and then you may
20 have overexpression of your product in that particular
21 cell. And so we need to be cognizant of those aspects
www.transcriptionetc.com
119
1 as well.
2 There are also chemical or structure-driven
3 safety aspects, so off-target effects. I think we're
4 going to hear a little bit about that in two speakers
5 from now, with Dr. Joung, and then also, the
6 interaction with the immune system. I mentioned
7 briefly about AAV. We also know with other
8 oligonucleotides, other antibodies, there are just
9 unique class-specific effects that you need to take
10 into consideration.
11 As far as the platform, once you move to the
12 platform, again, you can look at the properties of the
13 delivery vehicle. So I've spoken quite a bit about the
14 specific AAV capsid and understanding how that capsid
15 delivers. The capsid is really what's delivering it to
16 the different cell types. And so whatever --
17 regardless of what you have inside that capsid, the
18 capsid itself is usually what's going to deliver it to
19 those particular cells. Or you could also think the
20 same thing with a lipid nanoparticle or an antibody as
21 you get a sense, if you're utilizing them as drug
www.transcriptionetc.com
120
1 delivery vehicles.
2 The other aspects from platforms are assays
3 for measuring. So one of the unique things that, as
4 you start to develop this, there are assays out there
5 for measuring caspids and specific antibodies against
6 the capsid. So how do you measure that?
7 If it's against a specific capsid, you should
8 be able to leverage those across many different studies
9 or even other things such as anti-antibodies. So if
10 you're coming in with a biologic and you develop those
11 antibodies, if it's the same antibody being utilized in
12 a different thing, I think you can leverage some of
13 that data going forward for those biodistribution.
14 So here's an example of biodistribution data
15 that we generated in cynomolgus monkeys, with one of
16 our capsids, AAV HSC15. So we did this in two
17 different dose. And we were able to collect a variety
18 of tissues. This data now, these data, we can then
19 cross-reference for other studies that were using AAV
20 HSC15. And it really helps accelerate some of those
21 preclinical testing that we would be looking into for
www.transcriptionetc.com
121
1 utilizing AAV HSC15.
2 So finally, I just wanted to sort of wrap up
3 with the last few slides. As we think about this,
4 different targets require both product-specific
5 nonclinical characterization, as well as platform. So
6 just the example here, if we're using an AAV HSC15, and
7 we have disease one, the AAV HSC, the capsid, is
8 exactly the same, whether we're going after disease one
9 or disease two, just to exemplify this. So we can
10 cross-reference by distribution. We can utilize capsid
11 assays. We can utilize a lot of these different
12 aspects, just thinking about moving that forward to
13 help accelerate that.
14 But the inside is very different. So we may
15 have different promotors. We may have different trans
16 genes. And that's the product-specific aspect that we
17 do have to generate some testing for in --
18 preclinically, and so really looking at establishing
19 biological plausibility. So this disease and that
20 disease will have different cell-based models, perhaps
21 different animal models that we still have to go
www.transcriptionetc.com
122
1 through and assess, dose selection, and then finally
2 the underlying safety margin.
3 And then I talked a little bit about this
4 before. But the class effects can be informative but
5 perhaps not sufficient for a total safety assessment.
6 But we can pull that information in to utilizing as we
7 think about these. So one of the things just to
8 highlight here is just, as that data is being collected
9 and we're seeing more and more of these technologies
10 generate more data, publishing these data, you can
11 start to think about, from an individualized
12 perspective, moving those forward and using that to
13 help accelerate some of these individualized type
14 approaches.
15 So finally, platform-specific considerations,
16 in terms of selecting a dose, platforms that are
17 amenable to repeat dosing, so basically small molecules
18 and oligonucleotides, antibodies, protein replacement -
19 - a lot of those you can start at a lower dose and
20 increasing those doses with a safety margin. It really
21 gives you that opportunity to move forward with that.
www.transcriptionetc.com
123
1 And I think that's similar to what that example in
2 Batten's disease they were able to utilize that.
3 Ideally, align it with a PD biomarker because
4 we heard about this already. A lot of these
5 individualized diseases just don't have an established
6 natural history. So ideally, you want to know what
7 your drug is doing inside the patient and making sure
8 it's engaging with the target and seeing that activity,
9 so anything that can be done to spend time on
10 understanding those pharmacodynamic markers.
11 And then there also is some guidance around
12 microdosing. So maybe there's a unique opportunity
13 around microdosing options to even accelerate these
14 even further, as we think about taking these
15 individualized therapies forward. Again, coming back,
16 just to summarize on this. Platforms that are not
17 amenable to repeat dosing, so AAV, a lot of the other
18 genetic medicines, it is a single administration
19 aspect.
20 So I think the bar is a little bit higher in
21 terms that for particularly in individualized therapy -
www.transcriptionetc.com
124
1 - it’s an N of one -- getting that first dose right is
2 going to be essential for that particular patient. And
3 so really, that's where understanding the likelihood of
4 clinical benefit, understanding your preclinical dose
5 modeling data, going either from an in vitro, if that's
6 what you have, or really, pulling whatever you can from
7 in vivo to help drive that first dose. And maybe
8 there's an opportunity here, based on taking into class
9 effects. If you have the safety margin for an
10 individualized therapy, is that something to consider
11 that, instead of starting at the minimal efficacious
12 dose for that one patient depending on the benefit, the
13 likely benefit could be there? Do you start at a much
14 higher dose but still within your safety margin?
15 And then one of the things I wanted to -- as I
16 was thinking about this is putting in aspect around
17 repurposing existing drugs. This could actually be the
18 fastest because you'll probably have a host of
19 information already available to you, with respect to
20 safety profile, with respect to PD activity, to then
21 move pretty rapidly into these individualized
www.transcriptionetc.com
125
1 therapies. I think this is more of a rare exception
2 though because a lot of these individualized therapies
3 are going to be based on DNA sequence mutation and very
4 personalized. There just aren't a lot of existing
5 drugs out there that can target that. But it is a
6 possibility.
7 In terms of challenges and possibilities to
8 streamline this, I think the product-specific
9 characteristics, I think, have to be determined every
10 time. I think that's sort of par for the course. But
11 are there opportunities to streamline?
12 So one of the things I was thinking about is,
13 is there a form to share precompetitive platform data?
14 We heard a little bit earlier about IP. IP is always
15 going to be a consideration. But are there
16 precompetitive platform data? So whether it's AAV,
17 biodistribution, if someone is working on someone else
18 -- another AAV and we already -- and someone else
19 already has that data, how can we get that to share
20 particularly for a very severe disease that needs
21 therapy very quickly?
www.transcriptionetc.com
126
1 Commercial assays, specific to the platform,
2 so if we have neutralizing antibodies against certain
3 AAVs, can we share those? Can we get these out there
4 quickly? And then one of the other aspects that came
5 up around common manufacturing assays to support a
6 platform. So whether it's triple transfection and
7 HEK293s or whether it's SF9, are there different
8 aspects that we can take advantage of?
9 And then finally, I just want to end on really
10 thinking about this microdosing. Are there guidance or
11 white papers on the application that we can take
12 advantage of existing guidelines or initiatives that
13 are already there that the FDA has made available to
14 us? So microdosing in support of dose escalation for
15 individualized therapies, can we -- is there a
16 possibility to expand that?
17 And then the last thing I just want to bring
18 up is -- and it was brought up a little bit earlier --
19 the possibility and the openness of the FDA for what
20 are called Interact meetings. We found these to be
21 extremely helpful. And I think for something like
www.transcriptionetc.com
127
1 individualized therapy, this could be an avenue to
2 really start engaging with the regulators very early
3 because you can get quite a bit of feedback before you
4 start all of your nonclinical studies.
5 And so I just wanted to end on that aspect of
6 the different approaches. And again, thank the FDA,
7 thank the audience for listening to me.
8 DR. SANDUJA: Thank you, Dr. Seymour, for a
9 very interesting and exciting presentation. I would
10 now like to welcome our next speaker, Dr. Malachi
11 Griffith. Dr. Griffith is currently an assistant
12 professor of medicine and the assistant director for
13 the McDonnell Genome Institute at Washington University
14 School of Medicine.
15 Dr. Griffith completed his Bachelor of Science
16 with honors in biochemistry and biology in 2002 at
17 University of Winnipeg, followed by additional formal
18 training in computer science. He worked as a molecular
19 biologist and then as a computational biologist during
20 2003 to 2004 before beginning a PhD in medical genetics
21 and bioinformatics at the University of British
www.transcriptionetc.com
128
1 Columbia. After his PhD, he joined Washington
2 University School of Medicine in 2011.
3 Dr. Griffith's research is focused on the
4 development of personalized medicine strategies for
5 cancer using genomics and informatics. His lab has
6 made substantial contributions to open-access resources
7 for cancer research. Recently, the development of
8 bioinformatics for immunogenomics has become a major
9 focus of his lab.
10 Dr. Griffith now has more than 14 years of
11 experience in the field of genomics, bioinformatics,
12 datamining, and cancer research. He has over 80
13 publications and has received numerous research awards
14 and has held several large grants, including an NIH
15 K99. So I would like to welcome Dr. Griffith.
16 BIOINFORMATICS TOOLS FOR DEVELOPMENT, ANALYSIS &
17 PRECLININCAL TESTING OF INDIVIDUALIZED THERAPEUTICS -
18 DR. MALACHI GRIFFITH
19
20 DR. GRIFFITH: Okay. Thank you for inviting
21 me to speak today. It's been a really interesting
www.transcriptionetc.com
129
1 morning so far. I’m going to talk about something
2 that's been mentioned in several of the previous
3 presentations of bioinformatics but much more in
4 passing. So we're going to dive a little bit deeper
5 into the bioinformatics aspect of all this right now.
6 And I'd like to start by making an argument
7 that neoantigen vaccines in particular are a really
8 nice exemplar for individualized therapeutic. So we
9 heard a really interesting comment and question from
10 the audience this morning about neoantigen vaccines.
11 And I think that they're a great exemplar for a couple
12 of reasons.
13 One is that they're, in some ways, the most
14 personalized or individualized example of therapy that
15 we see today. So what's depicted here is a very high-
16 level 10,000-foot view of a workflow for developing a
17 neoantigen vaccine for an individual patient where you
18 start with a piece of their tumor tissue, taken at
19 biopsy or in some cases from a surgery. And then you
20 sequence the whole genome or whole exome of that tumor
21 DNA and compare it to the normal DNA from that
www.transcriptionetc.com
130
1 individual, from blood usually.
2 And you would usually also sequence the
3 transcriptome of that individual and then do a very
4 complicated bioinformatics analysis, which I'm going to
5 dive into a little bit more deeply here, to identifying
6 variation that leads to unique amino acid changes in
7 the genome of the tumor. So these are peptide
8 sequences that are specific to the tumor cells. And
9 then you use knowledge of those peptides on how the
10 immune system works to try to develop a vaccine made up
11 of these peptides that will stimulate the person's
12 immune system to respond to their tumor.
13 And so a typical example of this right now, in
14 many of the clinical trials that are underway, is
15 something like 5 to 20 peptides that are unique to the
16 tumor cells that you manufacture in a variety of ways.
17 And there are various manufacturing strategies being
18 used that get delivered, in the hopes of stimulating
19 the immune system to attack their tumor. And those
20 peptides are totally unique to that individual.
21 They'll never be used again. You're leveraging the
www.transcriptionetc.com
131
1 unique features of that tumor genome.
2 They're largely passenger mutations, so they
3 don't need to be functional per se. They just need to
4 be present in every tumor cell and cause an amino acid
5 change. And because they're passenger mutations and
6 they don't drive the biology of the cancer, every
7 person has a somewhat random set of them.
8 So I've been involved in probably now 50 to
9 100 vaccine designs. We've never used the same peptide
10 twice. It's always a different set for every single
11 patient.
12 So it's sort of the pinnacle of individual or
13 personalized therapy. And the other sort of large
14 reason, I think, that it's a good exemplar is just the
15 scale of the target patient population. Since somatic
16 mutation is a fundamental feature of virtually all
17 cancers, the potential application is potentially all
18 cancer patients. Everyone has an adaptive immune
19 system, and everyone's tumor has unique features of
20 those cells that could be targeted by their adaptive
21 immune system.
www.transcriptionetc.com
132
1 So the number of people that could potentially
2 benefit from this strategy is huge. So you have a
3 combination of a challenge, which is that it's
4 completely personalized. There's never going to be an
5 off-the-shelf drug that you can give to someone. But
6 at the same time, the potential market is absolutely
7 massive.
8 So just to dive in a little bit more detail in
9 terms of what this pipeline looks like. So I don't
10 expect you to read this. It's just to -- along with
11 several slides, to give you kind of a sense of the
12 complexity and detail that goes into this.
13 So it all starts with sequence data. So we've
14 heard this morning how important access, robust access
15 to that data is. And there's a series of complicated
16 bioinformatic steps that are undergone to sort of
17 convert that in raw genomic information into
18 predictions of first variation.
19 So you align your sequences against a
20 reference genome, and then you perform a variety of
21 different kinds of genomic variant calling and HLA
www.transcriptionetc.com
133
1 typing to characterize the immune system of that
2 patient. And then you run those variations through
3 customized pipelines that attempt to prioritize and
4 rank these variants according to their potential
5 usefulness as an immunogenic target. So all of this is
6 incredibly complicated. And there are a ton of tools
7 out there. And it's becoming increasingly sort of
8 automated and robust. But it's still also quite an art
9 form.
10 So currently, the vaccine design process
11 involves considering quite a large number of factors.
12 And this is often done in an immunotherapy tumor board
13 setting for the trials that are experimenting with this
14 therapeutic modality. And this is a group of experts
15 in immunology, genomics, bioinformatics, and the
16 treatment of the type of cancer that's being targeted,
17 who consider a variety of criteria for each of the
18 candidates that you're thinking of making the
19 individual personalized vaccine for.
20 And these are some of the things that are
21 features of the patient, and some of them are of their
www.transcriptionetc.com
134
1 genome. And some of them are individual to the
2 particular variants that you're thinking of targeting.
3 And I've just listed some of the examples on the right
4 side. And really, what we're doing in my group is
5 trying to automate as much of this as possible,
6 formalize the stuff that still involves human
7 intervention, replace as much of the human interaction
8 with machine learning as possible, and develop sort of
9 best practices and SOPs to help make this process
10 really reproduceable.
11 So we've heard a lot this morning about
12 process. This is a great example of that. So the
13 therapy is always different, but the process really,
14 really matters. And it's an incredibly complicated
15 process. So it's right for people thinking deeply
16 about the details of it and how we can make it robust
17 and reproduceable and make the process accessible to as
18 many patients as possible.
19 So part of the reason why this is a moving
20 target and we're still developing new approaches is
21 that this is a very new treatment strategy. And we're
www.transcriptionetc.com
135
1 just still kind of learning the rules of what makes a
2 good immunogenic neoantigen, so the peptides that are
3 specific to a tumor that actually lend themselves to
4 this approach. And one of the best ways that we're
5 learning this is by doing it in early-stage trials in
6 patients. And I've listed the trials that I'm involved
7 in here, which are a relatively small set of the trials
8 that are underway worldwide.
9 As you can see, they involve a variety of
10 cancer types, and they’re relatively small scale being
11 mostly phase one trials. But they all follow the same
12 process, which is that we start with a patient sample,
13 a piece of their tumor. We sequence the genome of it,
14 and we go through a pipeline that is both bioinformatic
15 and manufacturing and a preclinical assessment to
16 arrive at a vaccine that is actually safe and able to
17 be delivered to the patient by injection either
18 peptides or a dendritic cell vaccine or, in some cases,
19 the vaccine is incorporated into vectors that are sort
20 of similar to some of the vectors that have been
21 described this morning for delivering genetic payloads.
www.transcriptionetc.com
136
1 There's a huge amount of variability in these
2 trials. So they vary in their delivery approach. As I
3 mentioned, there's several manufacturing and delivery
4 strategies being used, the site of the tumors that are
5 being considered. It's almost every cancer type now
6 that has one of these trials underway somewhere. These
7 are some of the examples of the ones that are happening
8 locally at Wash U.
9 And then they vary in their combination with
10 other therapies. So many of them are being combined
11 with checkpoint blockade therapy, where the idea here
12 is to sort of take the brakes off the immune system,
13 while also simultaneously giving a kind of roadmap to
14 the immune system, so telling the immune system, these
15 are the unique features of the tumor in this patient
16 that you should be targeting, and then simultaneously
17 sort of stimulating the immune system to attack. So in
18 many of the talks we heard this morning, there was
19 concern about sort of immune toxicity or the effects of
20 the immune system are almost a problem.
21 But here, it's the reverse. We're trying to
www.transcriptionetc.com
137
1 enrage the immune system against the tumor but also
2 sort of provide some guidance in terms of what
3 specifically it attacks. And the level of specificity
4 here is quite exquisite because it's this very
5 personalized genomic base therapy individualized to
6 each patient.
7 So what I'm going to do now is just walk
8 through a list of some of the bioinformatics concepts.
9 So these are the things that keep me up at night, when
10 we're thinking about how do we do this whole process
11 from raw data to vaccine design and production in a
12 reproduceable and robust way so that the process is
13 constrained, and given the same set of input data, you
14 would arrive at the same answer. And I'm not going to
15 list those here because I'm going to go through them
16 one by one.
17 So it all starts with the sequence data, which
18 must ideally be of high quality. This part of the
19 process has become quite robust and sort of production-
20 ified, if you will, where there are many, many
21 sequencing cores and facilities and services where you
www.transcriptionetc.com
138
1 can send nucleic acid for the tumor and for the
2 reference blood and get DNA and RNA isolated, libraries
3 constructed, sequences generated. And then you think
4 of this sort of commodity sequencing, where everyone
5 has sequencing from the same platform, and they kind of
6 all look the same. The reads will be a certain length,
7 and they'll be paired and so forth. Those are certain
8 standards that have been widely adopted.
9 And I think that's largely true. But I would
10 just caution at this stage that the apparent
11 consolidation of many sequencing efforts to single
12 sequencing platform can lead to a false sense of
13 reliability in terms of sequence quality and the nature
14 of each sequence dataset. So I think that assessing
15 quality of your raw sequence data is still very, very
16 important. But I won't go into much more detail on
17 that.
18 The analysis of the data really starts with a
19 reference genome. So people think of the human
20 reference genome project as complete. It's not really
21 complete. It's still ongoing. But there is this sort
www.transcriptionetc.com
139
1 of pervasive fallacy that there's one human reference
2 genome, which for a bioinformatician is very much not
3 true.
4 So while the raw human genome assembly is
5 centrally maintained and you can sort of go and get the
6 centrally maintained build or assembly, there are many,
7 many, many derivatives of it. And these are in common,
8 common use. So we all hopefully know that there are
9 multiple assemblies or builds in common use. So we
10 think of build 37, build 38, or HG19, HG20. But there
11 are also many subversions, patches within those. And
12 the actual raw files used in a referenced genome can
13 matter.
14 And so usually, what's happening is people
15 aren't getting the raw assembly files themselves.
16 They're getting them through some second party, like
17 Ensembl, UCSC, 1000 Genomes, the Genome Data Commons.
18 And each of these vary in a variety of ways that can
19 really matter for downstream analysis and
20 interpretation. So for example, some of them used so-
21 called decoy sequences. Some of them include
www.transcriptionetc.com
140
1 alternative haplotypes and some don't.
2 There are light versions that simplify the
3 genome down to the chromosomes and throw away all of
4 the unplaced contigs. Chromosome naming amazingly is
5 not consistent to this day. Some viral genomes are
6 included, and some referenced genome sequences repeat,
7 masking may vary, et cetera.
8 So this still remains a large problem in terms
9 of consistency. The Global Alliance is really working
10 on trying to standardize some of this by developing
11 SOPs and best practices for uniquely identifying what
12 your reference genome really is. Variant discovery is
13 another area of a lot of variability. So you start
14 with alignment. And then with an alignment, you run a
15 series of variant calling algorithms or transcriptome
16 analysis.
17 And there is a huge diversity of how these
18 algorithms work and how they're used in combination.
19 And there are generally a series of tools used for each
20 broad type of variation. So for example, you might
21 have three variant callers, strelka, mutect, and
www.transcriptionetc.com
141
1 varscan, just to pick some examples, for calling single
2 nucleotide variants in a different set of variant
3 callers for SVs. And so you could have dozens of tools
4 potentially involve, just in identifying the potential
5 variation that's the grist for the mill, to identify
6 your neoantigens for these personalized vaccines.
7 Because of the complexity of this process,
8 again, manual review human intervention remains a
9 common part of many of these pipelines, where humans
10 actually look at raw sequence data. They manually
11 review variant calls that they're going to invest a lot
12 of effort or make something around. And so that part
13 of the process can be variable. We've worked to
14 develop standard operating procedures for the manual
15 review aspect of this and, also, machine-learning tools
16 to help automate it so to take humans out of the
17 equation.
18 Once you have variance, so you have identified
19 a genome variation that could be a source of a
20 neoantigen, there's a big representation problem in the
21 field. We can't agree on how to refer to variance.
www.transcriptionetc.com
142
1 And this creates a lot of confusion and inconsistency
2 across the field.
3 So I'm showing at the top here six different
4 depictions of the same variants, all being named in
5 different ways, only one of which is actually
6 unambiguous and computationally interpretable, and even
7 then is not in a very efficient computationally
8 interpretable form. So that's the bottom one. And the
9 bullet point is an HGVS string. The others are just
10 essentially colloquial ways of referring to a variant -
11 - that many of us know what BRAF V600E means. But
12 that's a sort of ambiguous way of representing it, to
13 just name it like that.
14 So this creates a problem because it's
15 difficult to know when two resources or two groups or
16 two labs are talking about the same or different
17 variation, which creates a sort of consistency
18 challenge. Again, there are many efforts to harmonize
19 variant identification. I'm involved in several of
20 these consortia that are sort of having conversations
21 around how we develop standards and ontologies to
www.transcriptionetc.com
143
1 really fix this part of the problem.
2 And the one that I really like to mention
3 specifically is the ClinGen Allele Registry, which is
4 depicted on the right here. And this is not the
5 perfect solution, but it's available now. And it fixes
6 a lot of the variant identification problems for many
7 people.
8 Just like the reference genome has many
9 versions, the reference transcriptome has many
10 competing versions. And when you're trying to think of
11 an amino acid change that arises from a change in the
12 tumor genome, that relies on interpretation in the
13 context of a specific transcript sequence. And we
14 don't actually know what that is. It's an
15 interpretation. And so it relies on predictions for
16 what transcripts look like.
17 And there are many competing reference
18 transcriptome efforts. Each of them handle prominence
19 and versioning differently, which creates a problem for
20 building robust reproduceable pipelines. For example,
21 Ensembl versions each individual transcript and their
www.transcriptionetc.com
144
1 overall set. But they're currently on release 99,
2 which creates another problem, which is that it's
3 constantly shifting targets.
4 You always have this temptation to be on the
5 latest set of reference transcripts, but at the same
6 time, that creates an instability in your pipeline
7 where it makes it harder to kind of lock down the
8 pipeline, where you can run it multiple times and get
9 the same answer. But Ensembl is popular among
10 bioinformaticians because of their sort of formal
11 handling of certain aspects of this problem. But it's
12 really important.
13 So we've identified several examples in our
14 own recent experience in these clinical trials where,
15 for a given referenced transcriptome, how you interpret
16 genomic events can really radically change based on how
17 you prioritize transcripts and consequences. So
18 there's this one variant, one consequence problem. We
19 like to think of one variation in the genome leading to
20 one amino acid change in one gene. But it's more
21 complicated than that because of overlapping genes,
www.transcriptionetc.com
145
1 because of alternative isoforms of genes.
2 And so there's very complicated tools that
3 attempt to solve this problem for you. And they have
4 to make difficult choices about how to prioritize the
5 referenced transcript that you're using for your
6 inference. And this can create a variety of problems.
7 So two examples of problems we had recently where cases
8 where the variant effect predictor we were using
9 changed the way they internally represented frame
10 shifts. And this caused us to have the potential to
11 create incorrect peptide sequences. And we can also
12 miss high-priority targets when an inappropriate
13 transcript gets selected as the highest priority.
14 Clinical variant and gene interpretation is
15 another challenge that several consortia are now
16 working to resolve. So the -- in the sort of rare
17 human disease space, the ACMG guidelines have really
18 helped to solidify how we do variant interpretation.
19 In cancer, there's been -- we're sort of behind because
20 we have the somatic variant problem that's been less
21 addressed. But we're starting to make progress with
www.transcriptionetc.com
146
1 the AMP, ASCO, CAP guidelines and other efforts like
2 CVC. And again, the Global Alliance has really been
3 organizing efforts to improve the consistency of both
4 gene and variant interpretation in a clinical context.
5 And then I really just want to end on the sort
6 of most bioinformatics heavy aspect of this whole
7 process, which is that you have this incredibly
8 complicated analysis pipeline for individualized
9 therapeutics. I'm depicting a small slice of our
10 pipeline here as a graph, with nodes and edges,
11 depicting steps, tools, data coming in, and
12 interpretations coming out. That's actually a small
13 piece of the overall neoantigen vaccine prediction
14 pipeline that we use that goes from raw data to a new
15 antigen vaccine. It involves dozens of tools, hundreds
16 of parameter settings, hundreds of input and output
17 files, and thousands of individual compute steps, which
18 makes it very difficult to have an actual reproduceable
19 pipeline.
20 So it gives us this question of how do we
21 actually ensure reproduceable results for very complex
www.transcriptionetc.com
147
1 pipelines? And so this is an area that we've really
2 been working on formalizing. And the short answer is
3 you should adopt a formal way of describing your
4 pipeline first of all. So there's been several recent
5 advances in things like workflow definition languages.
6 You should containerize everything. So place
7 all of tools inside of containers that isolate the
8 environment from compute dependencies. And you should
9 use a workflow execution system that runs the whole
10 pipeline.
11 And then ideally, you should organize these
12 layers into an analysis platform. And there are some
13 great examples out there like TARA. And don't forget
14 the importance of software engineering and
15 bioinformatic support.
16 So it's harder than it seems to keep a
17 computational pipeline locked down. There's an
18 assumption among many that because a bioinformatics
19 pipeline is computational, it must be inherently stable
20 or reproduceable, and this is actually surprisingly not
21 true.
www.transcriptionetc.com
148
1 It's potentially stable and reproduceable.
2 But that's actually much harder to achieve than you
3 imagine because of the complexity and the number of
4 steps and the dependencies on environment and how much
5 those environments and inputs and reference files --
6 everything can change. And it's harder than it sounds
7 to keep things locked down. So I'd just sort of urge
8 caution when thinking about the reproducibility of
9 these pipelines.
10 And then I'll just end with a short list of --
11 so I've thrown a lot of terminology and resources and
12 efforts out, but this is sort of the short list of
13 things, that if you're just going to check out a few
14 things or you want to become engaged in this kind of
15 work, these are the four things that I would recommend
16 starting with. I'd be curious to hear if anyone has
17 any examples of bioinformatics issues for
18 individualized therapeutics that I really missed. And
19 then I'll just end by acknowledging the wonderful group
20 that I'm privileged to codirect with my twin brother
21 and partner in crime, Obi, and of course, my funding
www.transcriptionetc.com
149
1 from the NHGRI and NCI in incredibly grateful for. I
2 look forward to talking to you later if you have
3 questions.
4 DR. SANDUJA: Thank you, Dr. Griffith. So
5 we'll now move on to our next speaker, Dr. Keith Joung.
6 Dr. Joung is currently a Desmond and Ann Heathwood
7 Research scholar and pathologist at Mass General
8 Hospital and a professor of pathology at Harvard
9 Medical School. Dr. Joung holds a PhD degree in
10 genetics from Harvard University and an MD from Harvard
11 Medical School. He's a leading innovator in the field
12 of gene editing.
13 Dr. Joung has pioneered development of
14 important technologies for targeted genome editing and
15 epigenetic editing of human cells. He has received
16 numerous awards, including an NIH Director's Pioneer
17 Award, an NIH Director's Transformative Research
18 Project R01 Award, the MGH Research Scholar Award, and
19 an NIH R35 MIRA Award. So we would like to welcome Dr.
20 Joung.
21
www.transcriptionetc.com
150
1 DEFINING OFF-TARGET EFFECTS OF GENE EDITING
2 TECHNOLOGIES - DR. J. KEITH JOUNG
3
4 DR. JOUNG: Thank you. And thank you to Dr.
5 Marks and the other FDA staff for the opportunity to
6 come and speak. So I'm going to talk today about
7 defining off-target mutations and effects of gene
8 editing technologies broadly. I have a conflict of
9 interest slide, which I'm required to show you by my
10 institution.
11 And so what I hope to do today is three
12 things. One is review the challenges and strategies
13 that exist now for defining gene editing nuclease, as
14 well as sort of next-generation CRISPR-based editor
15 off-target effects, and then tell you about our latest
16 assay, which we call ONE-Seq, which is a universal
17 platform for identifying off-target effects of all gene
18 editing nucleases and CRISPR-based editors that we
19 think also has advantages of scalability as well as
20 reproducibility, and then, at the end, just share some
21 perspectives on kind of the state of the field and then
www.transcriptionetc.com
151
1 what I see as issues and challenges moving forward. So
2 first, starting with a review of the strategies and
3 challenges involved in defining off-target effects of
4 gene editing nucleases.
5 So as Dr. Marks mentioned in his intro remarks
6 this morning, there are now a variety of different
7 platforms for doing gene editing. There are the four
8 sort of classical, if you will, nuclease platforms,
9 zinc finger nucleases, TALENs, mega-nucleases, and
10 CRISPR CAS RNA-guided nucleases, and then more next-
11 generation technologies, such as the CRISPR-based
12 cytosine- and adenine-based editors developed by David
13 Liu's group, which used the CRISPR system to direct
14 specific nucleobase deaminase enzymes to specific
15 locations in the genome.
16 So we've gotten very good now at being able to
17 put mutations where we want in the genome. But one of
18 the big challenges for the field for many years has
19 been defining and quantifying where else in the genome
20 we may be making alterations other than our intended
21 on-target site. And part of the challenge is that the
www.transcriptionetc.com
152
1 alterations in use by nucleases and base editors are
2 double-strand breaks and nicks. And so these are
3 short-lived events that are then repaired by cellular
4 DNA repair processes.
5 And so it's hard to actually directly identify
6 these alterations. And instead, what we do is identify
7 their outcome, so things like nonhomologous mediated --
8 nonhomologous end joining mediated indels or the base
9 substitutions induced by base editors. This can be
10 challenging to do, particularly in repetitive genomic
11 regions. And because of limitations in our ability to
12 do sequencing, it can be hard to distinguish these
13 alterations relative to background mutations or just
14 errors in the process of doing the analysis itself.
15 And this is particularly true for base editors.
16 Another challenge is that there really is no
17 gold standard for the field for off-target
18 determination. Whole genome sequencing is neither
19 practical nor particularly sensitive for finding these
20 alterations. And then on top of that, because of
21 tremendous interest in the field, it's a rapidly
www.transcriptionetc.com
153
1 evolving space with continuous improvements and
2 discoveries coming at a very rapid pace.
3 So how does the field address this? Well,
4 essentially right now, we use a consensus two-step
5 approach for being able to identify gene editor off-
6 target mutations. So the first step is what's called a
7 nomination or discovery process. And here, what you're
8 trying to do is identify potential sites of off-target
9 cleavage or mutations in a surrogate setting.
10 And so that surrogate setting can either be
11 cell-based assays, like the GUIDE-Seq assay developed
12 from my group and bless/bliss methods developed by Feng
13 Zhang’s group. Or it can be in vitro methods, where
14 you in a test tube have purified genomic DNA and
15 purified nuclease or base editors, and you ask whether
16 you can identify off-target effects. Now, you want
17 this to be as sensitive a method as possible. And you
18 want it obviously to be genome-wide in scope, so that
19 you can define the superset, if you will, of all
20 possible sites that the editor might be making
21 alterations at.
www.transcriptionetc.com
154
1 It's also important to note that not all sites
2 identified in this step may ultimately be -- ultimately
3 show evidence of alterations in the context of the
4 cells that you actually want to modify for a
5 therapeutic because there are other factors that come
6 into play that would be specific to your therapeutic
7 setting, such as, for example, the epigenetic status of
8 the gene. So the second step then is to take all of
9 the sites that you get from the nomination process, and
10 then ideally in the setting, therapeutic setting of the
11 cells that you want to be able to the therapy in,
12 actually look directly at those sites and ask whether
13 or not you see evidence of alterations at those sites.
14 And so this is why it's really important for that first
15 step to be as sensitive as possible because, if you
16 don't identify it in that first step, you won't even
17 look at it in the second step.
18 And here, there are a number of challenges,
19 which I'm not going to read through, but that also
20 exist with identifying whether an alteration has
21 occurred here. Typically, what people have been using
www.transcriptionetc.com
155
1 is targeted amplicon sequencing to be able to look for
2 the presence of indels. But it is important to note
3 that it is equally important to try to identify other
4 types of alterations that can occur when you create
5 more than one double-strand break in the cells, so
6 things like large-scale inversions or deletions or
7 translocations.
8 And then overall, I think it's important to
9 try to quantify risk as the sum, if you will, of the
10 on-target effect, but also all of the off-target
11 effects and the overall double-strand break burden that
12 you're actually inducing in the cell. Now, there's a
13 number of critical parameters. I don't have time to go
14 through all of them in a talk this short. But I do
15 want to mention assay sensitivity as being one that's
16 particularly important.
17 As I mentioned, for the nomination or
18 discovery step, this is very important because, if
19 you're not sensitive enough to pick up everything,
20 again, you won't even bother to look at it in the
21 second confirmation step. And in vitro assays, I
www.transcriptionetc.com
156
1 think, have the advantage over cell-based assays for
2 the nomination step of being more sensitive. We can't
3 quantify that sensitivity at this point, again, because
4 of the low frequency sites having some challenges being
5 able to confirm due to the error rate of next-gen
6 sequencing, which also limits the second confirmation
7 validation step as well.
8 Because if you amplify a section segment of
9 the genome, the process of doing PCR and then doing
10 next-generation sequencing, you can often see indels at
11 those sites at rates anywhere from 0.1 to 0.01 percent.
12 And so distinguishing a real alteration introduced by
13 your base editor or nuclease of interest from just
14 background error rates can be challenging. And then
15 ultimately, too, risk assessment is important because
16 you may make a break, but it may not have any --
17 ultimately any functional consequence. But here, we're
18 limited to some degree, or to a large degree, by our
19 understanding or knowledge of biology and genome
20 function at a particular identified off-target site.
21 Assay quality control is also equally
www.transcriptionetc.com
157
1 important. So some of the parameters to think about
2 here are positive and negative controls, particularly
3 for a negative study that shows or reports no
4 detectable off-target effects. It's also important to
5 account for sequence variation relative to reference
6 genome sequence.
7 And we've just heard about some of the
8 challenges, even in what constitutes referenced genome
9 sequencing in and of itself. But certainly,
10 differences among different cell types are important to
11 account for, so doing an untreated control is very,
12 very important when you're looking for off-targets.
13 And then there's a lot of other parameters, which
14 again, I don't have time to get into, but that relate
15 to the number of input genomes that are going into your
16 assay, the number of assays that you do, and assay
17 replicates that you do and biological replicates that
18 you do, sequencing depth, and then all the informatic
19 pipelines that you use to actually process the data.
20 So there are a lot of challenges. The NIST
21 has formed a consortium on gene editing led by Samantha
www.transcriptionetc.com
158
1 Maragh that I think is trying to address some of these
2 challenges. And it may be that, ultimately, commercial
3 service providers may be able to help with providing a
4 standardized trusted set of services that would address
5 these requirements.
6 Okay. So next, I want to tell you about a new
7 assay that has been recently developed by my lab. And
8 again, this is unpublished work. And here, it's an
9 assay we call ONE-Seq, which we think provides -- we
10 believe provides a universal platform for being able to
11 identify off-target mutations of gene editors both of
12 the nuclease class and the base editor class and to do
13 so with unsurpassed sensitivity.
14 So this is an in vitro assay, again, purified
15 components, in a test tube. If you look at all the
16 other assays that have been described previously in
17 this field, so things like Digenome-seq, SITE-seq, and
18 even CIRCLE-seq previously described by my lab, what
19 those assays do is they purify genomic DNA out of cells
20 or a particular tissue. And then they build some kind
21 of library out of that, treat with the nuclease, and
www.transcriptionetc.com
159
1 then attempt to identify in the context of all these
2 genomic sites, which are the sites that are actually
3 being cleaved by the nuclease.
4 And so the representation in the middle is
5 sort of the mess of genomic DNA, if you will, that you
6 get. In the human genome, obviously, you're going to
7 have at least three times 10 to the 9 of different
8 sites. And only a very small number of these will
9 actually end up being cleaved by the nuclease of
10 interest.
11 And what we've learned from a variety of
12 different studies performed to date is that these sites
13 all have some degree of resemblance to the intended on-
14 target site of the nuclease or the base editor. That
15 is that they are the same, but they differ at a certain
16 number of positions, as many as six or seven, within
17 the target site that you're trying to hit. And so it's
18 important then to emphasize that this is a very, very
19 small number of sites relative to the total content, if
20 you will, in a genome sequence.
21 So with ONE-Seq, what we do is we take whole
www.transcriptionetc.com
160
1 genome sequence for a particular cell type -- so for
2 example, for a human genome, we may use human reference
3 genome sequence -- and we go through and
4 computationally identify all the sites that have a
5 certain number of mismatches relative to the on-target
6 site. So you can go through and computationally
7 identify these. And you end up, for example, with
8 Cas9, typically, you have a list of sites of anywhere
9 from about 20,000 to 80,000 sites, depending on the
10 degree of orthogonality relative to the human genome
11 sequence.
12 And so you can then extract these sites out of
13 the genome and then synthesize all of them using high
14 throughput oligonucleotides synthesis, which now has
15 the capability to be able to synthesize up to millions
16 of these sequences precisely and give you exactly what
17 you want. And so what we do is we embed these
18 sequences in a fixed-length oligonucleotide. They're
19 always at the same position in the middle. And all of
20 these oligos are the same length.
21 And then after synthesis, they can be released
www.transcriptionetc.com
161
1 from the chip, converted to double-strand DNA. And
2 then this becomes your library that you then treat with
3 your nuclease of interest. And then I'll show in a
4 second how we extract the off-target cleavage sites.
5 So this type of approach using, if you will, a
6 more focused library that's been synthesized in vitro
7 relative to just using genomic DNA, has a number of
8 advantages. So one of them is that you can
9 characterize the entire library when you build it, just
10 by doing a simple MiSeq run because you're talking
11 about, again, anywhere from 20,000 to 80,000 sites.
12 And so you can sequence that library at high coverage
13 and know exactly what's going into the reaction in the
14 first place.
15 And so these are some examples of multiple
16 libraries that we've built for different target sites.
17 And you can see that the dropout of sites is actually
18 very, very low. It's also very, very reproduceable,
19 these libraries. So you do two independent syntheses,
20 and then you do sequencing and compare them. The
21 reproducibility of this is very, very high. And so
www.transcriptionetc.com
162
1 these are two different libraries here, which we've
2 done in duplicate and then done high-coverage
3 sequencing. And you can see that the reproducibility
4 between the libraries is also very high.
5 So this is a description of how we then use
6 this approach with Cas9 nuclease. So you take this
7 library, where you have a bunch of these different
8 target sites. You then treat with the Cas9 nuclease.
9 And so all the sites that are cleaved will be broken
10 into two. And then those free ends then serve as
11 substrate for ligating a sequencing adapter. And then
12 you can sequence the products that come out of this and
13 know which sites are being cut.
14 We then process the data. And so you end up
15 with these types of outputs, where you identify a whole
16 bunch of different sites. And typically, the highest
17 site, although not always, is the on-target site, shown
18 here with an asterisk. And you can assign what we call
19 a ONE-Seq score, which represents quantification of how
20 frequently these sites are being cut in the in vitro
21 reaction.
www.transcriptionetc.com
163
1 Now, another nice thing about this assay is
2 the unique capability to be able to set false-positive
3 thresholds for the assay. This is something that you
4 can't really do with a genomic DNA library. And the
5 way that you do this is for a given target site
6 library, let's say against a target in the FANCF gene,
7 instead of treating that library with a FANCF-targeted
8 nuclease, you target it with a different nuclease
9 target it to a different gene, so for example, to an
10 EMX1 target site.
11 And so what that allows you to do then is see
12 what the false positive rate is because nothing in that
13 library should be cut by a nuclease that doesn't target
14 that site. And so you can set precise cutoffs. And so
15 you can see the scatter plot on the right here are two
16 different library experiments. And again, you see high
17 reproducibility in the ONE-Seq scores of the different
18 sites. And the red lines represent the false positive
19 cutoffs that can be set based on doing these types of
20 mismatched nuclease experiments.
21 Okay. So to show you that ONE-Seq performs at
www.transcriptionetc.com
164
1 least as well, if not actually better than all the
2 other previous existing methods, here are comparisons
3 for four different sites of ONE-Seq against our GUIDE-
4 Seq cell-based method. And this is for identifying
5 bona fide, verified cleavage sites that actually are
6 cut in human cells. And so you can see that GUIDE-Seq
7 identifies -- sorry, ONE-Seq identifies all of the
8 GUIDE-Seq sites but then also identifies additional
9 sites as well.
10 This is a comparison of how well CIRCLE-Seq,
11 another method we had previously described, an in vitro
12 method we previously described, performs at finding
13 these bona fide GUIDE-Seq sites. And you can see the
14 CIRCLE-Seq sometimes can miss the different sites. The
15 two different colors represent doing CIRCLE-Seq on two
16 different cell type DNAs. And then finally, Digenome-
17 Seq, which is another in vitro method previously
18 described by Jin-Soo Kim’s lab, you can see also
19 misses sites, bona fide GUIDE-Seq sites. Although,
20 this was a result that's sort of known from the
21 previous literature.
www.transcriptionetc.com
165
1 We wanted to show that ONE-Seq was actually
2 capable of identifying sites in an in vivo context as
3 well. And so here, we used a liver-humanized mouse
4 model. And so this is a mouse model, where
5 essentially, a good portion of the mouse liver has been
6 replaced by human hepatocytes.
7 And so the nice thing about this model is it
8 allows you to examine off-targets in the context of
9 human cells but in a mouse model. And so this
10 addresses something that actually Dr. Marks was talking
11 about this morning, that when you want to look at off-
12 targets in a mouse model that's not really relevant to
13 look at off-targets in a mouse genome if ultimately the
14 goal is to use these nucleases for human therapeutic.
15 So I'm not going to go into the details of how we build
16 these types of mice. And I should say that this is
17 work done with Karin Musunuru at UPenn in collaboration
18 with his group.
19 So we wanted to do a particularly challenging
20 site. So we -- Karin had identified this GUIDE RNA for
21 Cas9 that targets early exon in the PCSK9 gene. And
www.transcriptionetc.com
166
1 the nice thing about this GUIDE is that it has very few
2 closely matched sites in the genome. So it has no off
3 by one, off by two, or off by three sites. So the most
4 closely matched sites are off by four and then go up
5 from there.
6 And so we like this site because we thought it
7 would be particularly challenging to find off-targets
8 for this type of site, and we wanted to see how ONE-Seq
9 would do in identifying potential sites that ultimately
10 would be modified in vivo. And so this is the ONE-Seq
11 output. The top site here would be the on-target site.
12 And then underneath are the off-target sites. Little
13 colored squares indicate mismatches relative to the on-
14 target site. And on the left column, there are the
15 ONE-Seq scores.
16 And so these are the top 40 sites, off-target
17 sites, identified by ONE-Seq. Karin then went and took
18 genomic DNA from the livers of these mice that have
19 been treated with this nuclease and asked whether you
20 could identify indels at those sites. So you can see
21 on the right that, the on-target site, you get very
www.transcriptionetc.com
167
1 high modification. And then within this set of 40
2 sites, there are at least four sites where we can see
3 very reproduceable, in triplicate, evidence of indel
4 mutations.
5 And these are sites that have either four
6 mismatches, the two higher sites, or five mismatches
7 relative to the on-target site. And you can see from
8 the numbers of sites there that it would be very hard
9 if you went through -- wanted to go through and
10 actually look at every single one of these sites in
11 these livers. But here, we're able to sort of rank-
12 order the sites based on ONE-Seq and then focus on
13 those sites and quickly identify off-target.
14 So this validates that ONE-Seq is capable of
15 finding these sites in this in vivo context and in this
16 more therapeutically relevant mouse model system as
17 well. I don't have time to go through all the data,
18 but we have a lot of data showing that ONE-Seq
19 outperforms other methods for other types of nucleases.
20 So Cas9 is a nuclease that leaves a blunt end. But
21 there are other nucleases that leave overhangs, like
www.transcriptionetc.com
168
1 Cas12a (Cpf1) or engineered zinc finger nucleases, mega
2 nucleases, or TALENs.
3 And here's some data showing you that ONE-Seq
4 outperforms GUIDE-Seq for identifying Cas12a nucleases.
5 So we find all the sites previously identified by
6 GUIDE-Seq. But we also find additional bona fide sites
7 using ONE-Seq. And we've also adapted ONE-Seq for the
8 base editor technologies as well.
9 And so here, we show that for a variety of
10 different sites -- that for the sizing-base editors,
11 ONE-Seq outperforms the Digenome-Seq assay, which is
12 the only assay that's been used to date to identify
13 off-target sites for cytosine-based editors. And it
14 also outperforms Digenome-Seq for adenine-based editors
15 as well. So we believe that ONE-Seq is, as I say, at
16 least as good, if not actually superior to all the
17 other methods out there for identifying off-target
18 sites for nucleases that leave blunt ends, overhang
19 ends, as well as for the cytosine- and adenine-base
20 editors.
21 Okay. So I have a few minutes left because,
www.transcriptionetc.com
169
1 actually, I think the intro was counted in my 20
2 minutes. So I think I have a couple more minutes to
3 talk about briefly some perspectives on looking
4 forward. So first of all, some viewpoints and
5 perspectives on the current situation. So it is not
6 possible to really ensure a complete lack of off-target
7 effects at present.
8 And I don't think it's a reasonable goal to
9 ensure that, given where our technologies lie right
10 now. The goal really should be to minimize off-targets
11 as much as possible. And this can be done through
12 things like protein engineering and other technologies
13 that have been described in the literature over the
14 last few years -- but still understand that they may
15 occur. And also, there needs to be an understanding
16 that not all off-target mutations will necessarily be
17 problematic.
18 We also need to recognize that there are
19 sensitivity limitations of the existing assays and that
20 there's also restrictions imposed by sampling for both
21 ex vivo and in vivo therapeutics. You just can't look
www.transcriptionetc.com
170
1 at all of the cells that are actually being modified.
2 Off-target profiling, it's also important to remember
3 is only one aspect of safety analysis. So there are
4 other tox studies that still need to be done. And in
5 general, when you talk about risk assessment of off-
6 targets, it's not possible to really make a general
7 recommendation about specificity outside of the given
8 intended use of a particular nuclease.
9 So looking forward, I do want to say that my
10 own personal opinion is that it is very, very important
11 that we continue to try to improve and extend both the
12 experimental and the computational approaches for
13 identifying off-target effects. There are a growing
14 number of voices now in the gene editing space,
15 particularly on the kind of academic research side but
16 also to some degree from the industry side, that some
17 folks will say, well, this is -- we've done enough.
18 We've done enough to be able to look at this.
19 And so I would strongly disagree with those
20 opinions. I think not continuing to do so, given the
21 limitations and sensitivity that existing technologies
www.transcriptionetc.com
171
1 have, and does a disservice to patients because these -
2 - especially as these technologies become more widely
3 used beyond a small number of diseases. At the same
4 time, I want to emphasize that I think it's important
5 that we not get hung up on trying to get a situation
6 where we have nothing that is -- we believe that no
7 off-target effects are happening.
8 So it's important for the sake of patients to
9 be able to define risk as well as we can at any given
10 moment and then to balance that against benefit. And I
11 continue to be very bullish and very, very enthusiastic
12 about trying to move forward with these different gene
13 editing therapeutics into the clinic. We do need to
14 continue to improve sensitivity as well as to be able
15 to have assays that predict functional consequences of
16 off-target effects. And I do think, ultimately, a lot
17 of the limitations in sensitivity we have come down to
18 error rates of next-generation sequencing technology.
19 So providing strong support for the development or
20 advancement of next-gen sequencing technologies and
21 improving their error rates will be very, very
www.transcriptionetc.com
172
1 important for the future as well.
2 There's other newer technologies. So David
3 Liu's group recently described prime editors. And so
4 we are working on trying to actually adapt ONE-Seq for
5 prime editors as well. And then there are other types
6 of, if you will, Cas9-independent or non-sequence
7 recognition-based edits that can occur. So for
8 example, we recently described RNA off-target edits
9 that occur with the cytosine- and adenine-based editors
10 that are due to the deaminase portion of those enzymes
11 functioning on their own. And other groups have
12 described off-targets on DNA that are not -- of base
13 editors that are not necessarily guided again by the
14 Cas9. And so these are important areas to continue to
15 push forward on as well.
16 And then another very important area that I
17 believe will be important to account for in the future
18 is human genetic variation. So at the end of the day,
19 these are sequence-specific reagents that are using to
20 target sites. And so the profiles of off-targets are
21 not going to be the same in everyone. You can define
www.transcriptionetc.com
173
1 them for a reference genome, but it's actually very
2 important, ultimately, to be able to define them for
3 specific individuals if not actually specific
4 populations or subpopulations of individuals.
5 And although it's not possible to do this
6 practically or feasibly for every single patient at
7 present, I think advancing technologies that allow us
8 to better understand the impact of human genetic
9 variation on off-target profiles is a very, very
10 important goal for the future. And ultimately, being
11 able to define off-target profiles individually for
12 patients is important as well. And then it's also
13 important to continue to focus research efforts on
14 better functional assays that allow us to identify what
15 the functional consequences are and, in particular, for
16 tumorigenic risks of off-target effects because, again,
17 this is an area where we just don't have very good
18 assays at present. And then in the longer-term,
19 follow-up studies and considering how to -- how and how
20 long to look at patients going forward are important
21 areas of continued development and research as well.
www.transcriptionetc.com
174
1 So I will stop there. I'm only like a minute-
2 and-a-half, I think actually overtime. I do want to
3 acknowledge Vikram Pattanayak and Karl Petri in my
4 group, who led the development of the ONE-Seq assay.
5 And this was largely funded by DARPA, with some
6 additional funding from NIH and the MGH Research
7 scholars, and done in collaboration with the Lie Lab,
8 so David Liu's group at Harvard and Greg Newby.
9 And then I also want to acknowledge the
10 American Society of Gene & Cell Therapy, who had some
11 input in some of the slides that I did early on for
12 another presentation I gave about a year-and-a-half ago
13 as well on this topic. So thank you for your
14 attention. And I guess I'll be happy to take any
15 questions about these issues in the Q&A. Thanks.
16 DR. SANDUJA: We'll start with our panel
17 discussion now. I'd like to welcome our speakers and
18 Dr. Zuben Sauna from OTAT to join us. Yes, please.
19
20 PANEL DISCUSSION WITH Q&A
21
www.transcriptionetc.com
175
1
2 MS. HOWARD: Hello. I'm Marilyn Howard from
3 the University of Pennsylvania. I have a question for
4 Dr. Griffith. I was very intrigued by the multiple
5 steps that you have in your bioinformatics process.
6 And I'm wondering what the metrics are for feeding back
7 to the machine learning in that process and whether or
8 not any of the clinical trials have yet reached a stage
9 where the clinical outcomes can feed back into the
10 machine learning.
11 DR. GRIFFITH: That is a great question. The
12 short answer is that it's pretty early days for the
13 clinical trial, so none of those clinical trials have
14 actually completed yet. And many of the trials involve
15 vaccination in a setting where the tumor is not
16 actually on board anymore, so they've been -- the tumor
17 has been surgically removed. And the vaccine is being
18 used in almost like a vaccine sense to prevent
19 recurrence. So it will take some time for the survival
20 information to accumulate. So we're probably at least
21 two or three years away from that sort of real gold
www.transcriptionetc.com
176
1 standard of is this actually improving outcomes.
2 That being said, you mentioned the word
3 metrics, which I love. There's a lot of other earlier-
4 stage things we can look at. So many of the trials are
5 building in a variety of innovative immunological
6 monitoring steps to try to get an early sense of are we
7 seeing t-cell responses, in terms of changes on the TCR
8 repertoire or functional assays of t-cells, screening
9 for particular candidates before and after vaccination.
10 So that data should come sooner.
11 It's still going to be relatively sparse
12 because the trials are so small. But we can probably
13 learn quite a lot on a per peptide basis because many
14 of the assays are giving us a specific readout and
15 useful metrics for each of the peptides that we tested.
16 In some of these patients, we have up to 20 candidate
17 peptides. And we really don't know what the rules are
18 for what makes a good peptide.
19 So we're potentially going to learn a lot in
20 the next few years, subject to sufficient sharing of
21 that data and all of the usual challenges that go into
www.transcriptionetc.com
177
1 these kinds of studies. But I think that there's
2 reason to be very optimistic that we'll improve our
3 understanding of how to actually design a vaccine
4 significantly in the next few years and hopefully, more
5 like four to five years for the gold standard survival
6 analysis.
7 MS. HOWARD: Thank you.
8 MS. ADOMAKO: Hi. My name is Jessica Adomako,
9 and I'm from Genentech. This question is for you. And
10 it's --
11 DR. SANDUJA: Can you please speak up? It's
12 hard to --
13 MS. ADOMAKO: Can you hear me?
14 DR. SANDUJA: Yes.
15 MS. ADOMAKO: Yeah. This question is a follow
16 on to what the previous person asked. And there's a
17 second part, which is to the FDA. You talked a lot
18 about the bioinformatic challenges. And I completely
19 agree with you. And my question is have you looked to
20 the work that has already been done by the community at
21 large in establishing standards for analytic validity,
www.transcriptionetc.com
178
1 everything from referenced genomes to databases to
2 validity of bioinformatic pipeline software, et cetera,
3 et cetera, in the related field of developing NGS-based
4 tests?
5 Because there's a lot that can be learned.
6 And it's the exact same question to the CBER folks on
7 the panel. Your compatriots at CDRH have really done
8 an amazing job of establishing widely accepted
9 community standards.
10 And what we would like to know is are these
11 learnings being shared? Are there things that you can
12 develop based on what they've already established? And
13 completely conceding that this field is, as you said,
14 constantly evolving, we're always chasing a new goal.
15 But can we ping off of what we've already done?
16 DR. GRIFFITH: Yeah. So that's a really great
17 point. And I meant to mention this actually in my talk
18 that I think that there are many, many common themes
19 and similarities between what is going on, for example,
20 in the new antigen vaccine design process that can
21 benefit from the years and years of labs producing,
www.transcriptionetc.com
179
1 conducting genomic diagnostic tests and producing a
2 clinical report with interpretations for a specific
3 variance. They face many of the same challenges.
4 So we absolutely can learn a lot from those
5 practices and standards. And we're definitely -- all
6 of the consortium that I mentioned really evolved out
7 of those groups and heavily involve experts that have
8 years and years of experience in those areas. So yes,
9 absolutely.
10 MS. ADOMAKO: I’m just waiting to hear from
11 someone from CBER. Are you speaking here?
12 DR. SAUNA: I didn't quite get what do you
13 want -- what is the question to CBER?
14 MS. ADOMAKO: The question to CBER is, has all
15 of the work that's gone on at CDRH in establishing
16 these protocols, processes -- they even have wonderful
17 final guidance documents that are now being widely used
18 by sponsors. Is any of that, do you think,
19 translatable to CBER?
20 DR. SAUNA: I think Dr. Carolyn Wilson wants
21 to address your question.
www.transcriptionetc.com
180
1 DR. WILSON: I'm going to just step in real
2 quickly. Carolyn Wilson from Center for Biologics. I
3 just wanted to say, yes, we actually have at the agency
4 level a genomics working group, where we've actually
5 been exchanging and learning from other centers,
6 working in this space for about five years. And then
7 we also have a CBER genomic working group. And we've
8 actually brought the CDRH guidance documents to that
9 group and looked at them very carefully, as well as ICH
10 guidance documents and so on. So certainly, we're
11 well-aware of those other efforts and are incorporating
12 them into our own thinking.
13 MS. ADOMAKO: Thank you.
14 MS. WITKOWSKY: Hi. Lea Witkowsky from
15 Innovative Genomics Institute at UC Berkley and UCSF.
16 That was a really great session. Thank you, everybody.
17 And I have a question for Keith Joung in particular.
18 Very exciting data.
19 I'm wondering for ONE-Seq, if you're starting
20 with a prediction algorithm, presumably, you're
21 starting that -- you’re running that prediction on some
www.transcriptionetc.com
181
1 sort of reference genome. And as we think about
2 individualized therapies and kind of harkening back to
3 the vaccines, developing these at an individual level,
4 how do you expect, or do you expect, to be able to
5 adapt that for individual patients?
6 And you mentioned a little bit about this
7 human variability. And what is the capacity, or where
8 do you see the future going to be able to run something
9 like that to catch things that might be off-targets
10 only for an individual that happens to have a mutation
11 that makes it a new off-target that you wouldn't catch
12 normally in a reference genome, for example?
13 DR. JOUNG: Yeah. That's a great question.
14 Thanks, Lea, for that. So I think one of the strengths
15 of ONE-Seq is the ability to be able to look in a
16 detailed way at specific sequence changes for an
17 individual, or potentially even a group of individuals.
18 It is dependent on having all genome sequence
19 data for that person. When we set out to start
20 developing ONE-Seq actually about two, two-and-a-half
21 years ago, we assumed two things would start to come
www.transcriptionetc.com
182
1 down in cost. So one would be oligonucleotide
2 synthesis, which I think has already begun to play out,
3 and hopefully will continue to only get less and less
4 expensive. And the other was the assumption in the
5 continuing drop in the cost of being able to do whole
6 genome sequencing.
7 So if you envision a world going forward where
8 it's relatively inexpensive to do whole genome
9 sequencing and to do oligosynthesis, then I think it
10 becomes very reasonable to assume that you could
11 combine those two, be able to practice ONE-Seq, and
12 then be able to get information that's specific to a
13 particular individual. I also think it's just
14 generally easier to scale something like ONE-Seq to
15 cover more people and variants in more people than it
16 is to scale something like doing one-off in vitro
17 assays, the way you do Digenome-Seq, CIRCLE-Seq, or
18 SITE-Seq right now.
19 So I hope that makes sense. And that is our
20 hope, is that we will be able to account for more
21 individualized genetic variability because, as I said,
www.transcriptionetc.com
183
1 at the end of the day, these things are sequence-
2 specific agents. And so they are -- you expect -- it's
3 entirely reasonable to expect that their impact will be
4 specific to a particular individual based on their
5 genome sequence.
6 MS. WITKOWSKY: Thanks.
7 MR. STEIN: My name is Aron Stein with Sangamo
8 Therapeutics. This is a question for Dr. Joung. This
9 is in regard to your methodology for the validation of
10 your targets using a humanized mouse. Why that model
11 versus primary human hepatocytes?
12 DR. JOUNG: You could do it certainly in
13 primary human hepatocytes as well. We validated ONE-
14 Seq-predicted sites in cells in culture. So there's no
15 reason why you couldn't do it that way. Although there
16 are some challenges with getting the reagents
17 efficiently into human hepatocytes. It's certainly not
18 unreasonable to try that experiment.
19 The reason we did it in the context to the
20 mouse was that we wanted to look at in vivo in an in
21 vivo setting, where you would be delivering these
www.transcriptionetc.com
184
1 reagents in an in vivo setting, whether we would
2 predict those off-targets there. Because if you look
3 at the literature, to date, no one has actually been
4 able to identify these off-target sites, especially
5 those that have a large number of mismatches relative
6 to the on-target site in the context of something like
7 an in vivo animal model. In fact, our -- especially
8 when you're using a nuclease that has been designed to
9 be relatively orthogonal to the human genome. And so
10 that's why we chose that particular guide because it
11 would be easy to spot an off by one, off by two, off by
12 three. That I don't think anybody would be surprised
13 by.
14 But this is the first demonstration to our
15 knowledge where you're able to find off by fours or off
16 by five sites in an in vivo animal model where you've
17 delivered the nucleases in that way. So there were a
18 number of reasons for doing it. You certainly could do
19 it in cells. I didn't mean to imply that you couldn't
20 do it that way.
21 DR. SEYMOUR: Okay. Thank you.
www.transcriptionetc.com
185
1 MR. ALDRICH: Time for one more? Okay. I
2 just wanted to follow up with Malachi on his answer to
3 a previous question. Regarding -- I think a lot of
4 folks don't quite appreciate that the biggest problem
5 that comes out of the predictive algorithm is false
6 positives for neoantigens that aren't really present or
7 -- and that that's a problem which we can address by
8 running spot tests against -- ELISpot spot tests
9 against t-cells and eliminating false positives by
10 identifying, of the predicted and synthesized top
11 candidate antigens, synthesizing the peptides that
12 correspond to the neoantigens and then testing them
13 against the patient's t-cells for reactivity.
14 We kind of eliminate that problem of the false
15 positives in the -- and in terms of issues that are
16 really front and center for me as a patient, it's
17 making sure that, of the 5 to 20 peptides that we're
18 going to use in a final vaccine, that they're all
19 validated in some sense as having a corresponding T-
20 cell, which, hopefully, they'll amplify. I just
21 wondered if you'd comment on that. Thanks.
www.transcriptionetc.com
186
1 DR. GRIFFITH: Yeah. I mean it's a really
2 interesting question and topic that I think both the
3 false positives and false negatives are interesting. I
4 think that we think of the false positives as being
5 more tractable because as you say, you're right. We
6 can think about validating them further, and we can
7 look for specific T-cell responses.
8 Although, I guess I would say that it's still
9 falling short of -- ideally, we would know that they
10 were not just T-cell immunogenic but therapeutically
11 useful. And that's sort of like the next stage of --
12 MR. ALDRICH: Yeah.
13 DR. GRIFFITH: -- another layer of false
14 positives that's yet to be learned about.
15 MR. ALDRICH: Right.
16 DR. GRIFFITH: But I think we also don't know
17 what we don't know, in terms of false negatives. So we
18 don't really have a great sense, of the candidates that
19 we're nominating, how many great candidates did we
20 leave on the table just for not knowing about them, not
21 looking for the right kinds of variation or
www.transcriptionetc.com
187
1 prioritizing them incorrectly because of our lack of
2 complete understanding of how the immune system works.
3 So I think that's also an area for significant
4 improvement. And we are starting to see a little bit
5 more by unbiased assays looking at with the peptide
6 mass spec elution dataset started to give you a bit of
7 a sense of just sort of serving, like what are all the
8 peptides that we're sticking to a particular MHC
9 molecule and getting a more comprehensive readout of
10 that. Although, it also has some pretty significant
11 caveats to those datasets that we could probably talk
12 about for an hour.
13 MR. ALDRICH: For -- we will. But just one
14 last thing to have you comment on is one of the things
15 that ties into the talk we heard earlier, about
16 platform versus product and characteristics of the
17 neoantigen platform, as opposed to the specific vaccine
18 -- one of the things that I'm very impressed by is
19 that, when you read across the literature, we have a
20 great deal of familiarity with peptide vaccines. I
21 mean, historically, it's something where there's a lot
www.transcriptionetc.com
188
1 of data, not necessarily neoantigen peptide vaccines,
2 but peptide therapies have been around a long time.
3 And I think that one of the problems or one of
4 the challenges is that we have to recognize when we
5 have a platform where we know enough, so that we can
6 declare the platform as relatively safe. And as a
7 consequence, if you perform best practices, with
8 respect to the neoantigen platform, you can be pretty
9 well-assured that a patient isn't going to drop dead
10 when they're -- in fact, one of the incredible things
11 about the neoantigen trials that are going on is that,
12 to my knowledge, there hasn't been a single severe
13 adverse event reported from a safety point of view.
14 And yet, we are still waiting for the first
15 approval of a neoantigen peptide vaccine for any
16 indication of cancer. And so one of the things that
17 occurs to me is that, gee, if we have a safe platform
18 and we know enough about it and it proves efficacious
19 in phase-one trials saving lives, what is -- where do
20 we get comfortable enough to say, oh, well, if you're
21 following this best practice in terms of the supply
www.transcriptionetc.com
189
1 chain, we really ought to make it available to lots of
2 cancer patients? Just that's where I'm coming from.
3 DR. GRIFFITH: Yeah. I mean I guess my main
4 comment would just be to completely agree with you. I
5 think the pieces are in place to create such a platform
6 or process that we can really carefully document and
7 become confident that it is robust and reproduceable
8 and safe. My impression is -- as I said, many of these
9 have not published. Basically, none of them have. But
10 the early impression does seem to be that the safety
11 profile of this approach is outstanding.
12 And so that comes back to my initial comment
13 that the potential patient population this could be
14 applied to is just huge. I mean if it's safe and even
15 just a little bit efficacious for some people and we
16 could do it cheaply and broadly, you could imagine this
17 being added into the course of treatment in so many
18 current clinical cancer regimes, just to get a little
19 boost, just to get a little boost from the immune
20 system against that person's tumor in a way that seems
21 to be very safe. But we need to establish efficacy, or
www.transcriptionetc.com
190
1 this is all moot. But yeah.
2 DR. SANDUJA: I would like to add another
3 point of view to that. From a regulatory perspective,
4 if you look at that, we have gained and we have
5 gathered a lot of confidence in the platform approaches
6 that these are able to really very confidently, and
7 with a lot of robustness, can inform safety of these
8 products. And that's what has enabled a much faster
9 translation into clinical trials.
10 However, when it comes to approval or
11 licensure, that's a different question there. So we
12 are gathering confidence with respect to safety of
13 these platform approaches to further facilitate. And
14 that's being evaluated as we move forward with clinical
15 development of these products.
16 MS. WALKER: Hi. Karen Walker from Genentech.
17 Thanks for some really interesting conversations. I
18 have a couple of questions, again, going back to the
19 data and the bioinformatics. While I agree that safety
20 is a very important aspect -- and so is efficacy, also
21 so is supply and control -- once you have -- you're
www.transcriptionetc.com
191
1 treating lots of patients and the variability that you
2 described increases exponentially, how do you continue
3 to learn and identify what is important out of all of
4 the data that you're gathering versus what is just a
5 datapoint?
6 And I think that's a really important question
7 to ask and to answer. So I would be interested in your
8 thoughts.
9 DR. GRIFFITH: Yeah. I mean it's a hard
10 question. Do you think the increased variability from
11 doing the set scale on sort of a population scale, is
12 the implication there because there are so many players
13 in the space, and that's kind of a wild west of a
14 hundred different people doing it a hundred different
15 ways? Or do you just -- or do you mean more that we're
16 just servicing the tip of the iceberg in terms of how
17 patients respond and how different tumors behave?
18 MS. WALKER: I think it's more the latter.
19 DR. GRIFFITH: Okay.
20 MS. WALKER: But I also think it's what you
21 mentioned about the standard reference genome changing,
www.transcriptionetc.com
192
1 and what are disease-causing elements or the disease-
2 related elements versus just natural variability in our
3 own genetic sequences. This is the kind of thing that
4 I think we need to understand more.
5 DR. GRIFFITH: Yeah. I mean I think from a
6 personalized genomics aspect of it, personalizing it to
7 the reference genome of the individual, so that you're
8 comparing that against their tumor, that is a tractable
9 problem and one that has been tackled in other areas of
10 cancer genomics and NGS testing for cancer with tumor
11 genomes. In terms of just, yeah, how do we figure out
12 how to do it better in the face of all this variability
13 for different cancer types, I mean, I guess the high-
14 level answer, I would say, is to share the data openly
15 and let people have access to it and try to make the --
16 look inside the black box, that we really need to
17 understand what's going on in these pipelines, what the
18 process actually is, and what features are important.
19 All of the metrics, the readout, everything
20 needs to be exposed, so that we can -- because it's so
21 complicated it's going to be very difficult for one
www.transcriptionetc.com
193
1 group to understand. The pipeline is already so
2 complex that it's actually pretty much impossible for a
3 single person to really even understand the whole thing
4 from end to end. So it's really going to be a sort of
5 community team science or big team approach, even just
6 for the bioinformatics part of it. And that's just one
7 small piece of the complexity of this overall process.
8 MS. WALKER: Thank you.
9 DR. SAUNA: Could I follow with the question?
10 DR. GRIFFITH: Yeah.
11 DR. SAUNA: So to follow up on this question,
12 would it help if -- so you're already at some level of
13 precision by looking at an individual. As you get to a
14 deeper level of precision, looking at particular cell
15 types or single -- if you do single-cell sequencing,
16 and particular subsets of that tumor, which would
17 probably be more susceptible to the antigen and making
18 a tumor antigen targeting -- say a metastatic cell, for
19 example, rather than the tumor cells in general. Would
20 that level of precision help making it more effective?
21 DR. GRIFFITH: I think it's definitely
www.transcriptionetc.com
194
1 relevant to interpreting the response to the therapy
2 because the tumors are heterogenous.
3
4 And this is definitely a consideration of
5 ours, when we think about designing the vaccine, do we
6 specifically target only those antigens that are
7 thought to be in the trunk or a clone of the tumor that
8 will be in all of the tumor cells and not in a
9 subclone? Or is it, in some cases, okay to target
10 subclonal mutations? And then in terms of interpreting
11 the immune response, then the single-cell analysis of
12 the tumor microenvironment becomes very, very relevant
13 and useful potentially. But it's also very much a
14 developing area where there's so much to be learned.
15 MS. MCLELLAN: Hi. My name is Lorraine
16 McLellan. I'm a cancer patient and actual cancer
17 survivor. But I have something that's going to relapse
18 here in the next year or two. And I have done my
19 genome, and I am hopeful that I can do an neoantigen
20 vaccine.
21 But here we are in the headquarters for the
www.transcriptionetc.com
195
1 FDA. And what I would like to ask, because each of you
2 have done such an impressive job, is if I was able to
3 grant you one wish each to give a message to the FDA
4 today about what you would like them to do near-term,
5 and near-term, let's say 12, 18 months, to advance your
6 work individually, what would that be?
7 Would that be a question that I could each of
8 you to answer, so that we have a real takeaway and some
9 action items? Because we do have a bit of a need for
10 speed. I can appreciate that it's going to take a
11 couple years to get to gold standard. But what does
12 the FDA need to do near-term for each of you to advance
13 your work? Thank you.
14 DR. GRIFFITH: Do you want me to start? So in
15 terms of my individual work, I guess I would rather
16 answer more for the overall translation of this work to
17 patients. I think there is a challenge. And I've
18 heard this from quite a few representatives of -- in
19 research and in industry, that it's difficult to think
20 about developing a commercial version of a new antigen
21 vaccine without any clarity around whether the process
www.transcriptionetc.com
196
1 can be approved as safe, something that we're
2 comfortable with.
3 So I think that this meeting is very timely.
4 And I assume it is accompanied by an interest on the
5 part of the FDA, although I have no idea, to gather
6 information about how one might think of ultimately
7 giving some kind of regulatory oversight to these -- to
8 the process that's come up so many times. So yeah, I
9 would encourage the sort of continued serious thought
10 of that idea.
11 Because right now, I think it's a little bit
12 of an impediment for someone who's thinking about
13 trying to do this as a company that there isn't really
14 an obvious -- you know, you're not going to be able to
15 patent a drug. So -- and there's sort of risk averse
16 problem, where if they're -- if you don't have any sort
17 of stamp of approval that the process has been
18 evaluated in some kind of formal way as being safe,
19 then it makes it seem riskier to pursue it.
20 DR. SEYMOUR: I’ll go second. So I think it's
21 a really good question and one of the things I've sort
www.transcriptionetc.com
197
1 of put a hat on in trying to determine those different
2 aspects. So there is getting these therapies to
3 patients in a timely matter. And I think it sounds
4 like that is of critical need because a lot of these
5 diseases do progress very rapidly. So you don't have a
6 lot of time, particularly for the individualized
7 therapies.
8 And I think some of the discussion topics that
9 the FDA is having right now, I think are a path towards
10 that. And so I look at what does it take to get
11 something to a patient so that you can test it rapidly.
12 I think a lot of the pathways are already in place for
13 that in getting it to go into early-stage testing very,
14 very rapidly, whether it's individualized therapies or
15 not.
16 The second aspect is moving towards the
17 commercialization and licensure. And I kind of divide
18 those two separately when really thinking about
19 individualized therapies versus, say, bringing
20 therapies for a much larger population. I think
21 they're two separate questions. And so I think my wish
www.transcriptionetc.com
198
1 to the FDA is to continue discussions like this, work
2 with both industry as well as academic partners to come
3 forward with creative ways of bringing and testing
4 these products safely in patients and always keeping
5 in mind -- and I think everyone does on that -- that
6 benefit/risk and the severity of the indication that
7 we're trying to move forward with.
8 DR. JOUNG: Yeah. I don't have any specific
9 request for the FDA in terms of the work that we do. I
10 think for the community, as a whole though for gene
11 editing, I would encourage the community to have maybe
12 broader and wider-ranging discussions about how to
13 better standardize how to benchmark and how to develop
14 consistency around some of the safety and in particular
15 off-target testing because I think it's become very
16 fragmented. And so some of the issues that I raised in
17 my talk I think are things that we as a community need
18 to address.
19 And if we can do that, I think it will make it
20 easier for companies, academics, to be able to know
21 what it is that they need to do. Also, that it is an
www.transcriptionetc.com
199
1 evolving and fast-moving field, so the standards will
2 change over time. So to the extent that we can build a
3 mechanism by which there is -- I don't know exactly
4 what the form will be or what the body would be that
5 would do this -- but some kind of ability to respond to
6 changes as they occur, to be able to keep those
7 standards current and up to date. I think that would
8 be helpful for the field.
9 DR. SANDUJA: And to conclude that discussion,
10 I would like to -- from the FDA side -- would like to
11 reiterate that FDA acknowledges the challenges that are
12 associated with development of these individualized
13 drug products. And as we have seen, during all our
14 presentations and also during the discussion, there are
15 pathways and there are opportunities to discuss these
16 challenges and come to an agreement how they can be
17 resolved. Of course, we already agree that the
18 standard paradigm of drug development may not apply to
19 development of these individualized therapies. And as
20 the science behind these products is continuously
21 immerging and evolving, the Agency itself is open to
www.transcriptionetc.com
200
1 these discussions and further efforts, like the one we
2 are having today, to continue development of these
3 products. Thank you.
4 DR. RAYCHAUDHURI: I would like to thank all
5 the speakers and panelists and the audience
6 participants for what has been really an excellent
7 discussion this morning. And I'd like to thank Dr.
8 Sanduja for moderating this session. We are now at the
9 lunch break. And we have a very exciting afternoon
10 session ahead of us. So I ask that you please return
11 back to this room at 1:15, and we will proceed with the
12 afternoon sessions. Thank you.
13 [BREAK]
14 SESSION 3: CLINICAL
15 DR. RAYCHAUDHURI:
16 So I hope everybody had a nice break. So in
17 the morning sessions, we heard about challenges and
18 opportunities related to manufacturing of gene
19 therapies and phage therapies and tools for safety
20 testing and development of individualized therapeutics
21 products. This afternoon we're going to focus on
www.transcriptionetc.com
201
1 clinical development. The challenges and opportunities
2 to leverage what is known to facilitate development of
3 related products. And in session four, we will come
4 full circle on the focus and the reason for this
5 workshop, which is to discuss how we collectively can
6 increase access for patients to these critical products
7 in a timely and sustainable way.
8 So we have a very exciting agenda set for this
9 afternoon. I would like to introduce Dr. Rebecca
10 Reindel. Dr. Reindel is a medical officer in the
11 Division of Vaccines and Related Products Applications
12 in the Office of Vaccines Research and Review at CBER.
13 Dr. Reindel will be the moderator for session three,
14 which is on challenges and opportunities for clinical
15 development of gene therapy and phage therapy products.
16 Dr. Reindel.
17
18 SESSION 3 MODERATOR INTRODUCTION: DR. REBECCA REINDEL
19
20 DR. REINDEL: Thank you. I'm really excited
21 to be part of this third session. We have two
www.transcriptionetc.com
202
1 presenters. Dr. Schooley and Dr. Kohn will be
2 presenting, and then we'll have a panel session to
3 follow similar to the other sessions.
4 So by now you've seen several versions of this
5 slide. And I wanted to bring it back around to
6 clinical development and talk about the clinical
7 development of individualized therapeutics within this
8 paradigm. And as you can see, this figure describes
9 clinical development all the way through discovery and
10 pre-clinical, which we've covered quite a bit this
11 morning, and into the clinical phase of things where I
12 think this afternoon's sessions will sort of start to
13 take over.
14 Within this paradigm, typically you see
15 smaller Phase 1 studies that are designed to assess
16 safety and dose selection and then move up through
17 larger populations into Phase 2 and Phase 3 studies,
18 which study safety in an ongoing manner, and also start
19 to assess for effectiveness. However, individualized
20 therapeutics may not lend themselves as well to this
21 classic paradigm. And therefore, it's really important
www.transcriptionetc.com
203
1 that we look for opportunities for flexibility within
2 this paradigm with regard to clinical development.
3 In this session, we hope to identify both the
4 challenges that we face and anticipate in this space,
5 as well as any opportunities we can identify for
6 ongoing clinical development of individualized
7 therapeutics. So I like to think that every challenge
8 presents us with an equal opportunity to meet those
9 challenges and rise above them. So this is no -- by no
10 means an exhaustive list of all the challenges we face
11 in this space but some key highlights that we hope to
12 get to today as well as others include the following:
13 clinical investigations in the context of potential
14 manufacturing challenges, many of which were discussed
15 this morning.
16 In these studies, we need to pay a lot of
17 attention to study design when the product may be
18 different for every recipient or may need to be
19 tailored to a specific recipient or subject in real
20 time, such as to accommodate individualized treatment
21 of a patient with an infection that requires a specific
www.transcriptionetc.com
204
1 phage for that infection and then the impact of these
2 differences on the study procedures in the trial and
3 the interpretation of data that arises from these
4 trials. How do we interpret data that comes from
5 individual patients or subjects within the context of a
6 clinical trial that may include only maybe one or
7 several subjects? And also, how are we to interpret
8 novel endpoints? For example, micrologic- --
9 microbiologic endpoints for bacteriophage treatment.
10 This may pose some challenges in interpreting endpoints
11 in the absence of precedent.
12 Again, so what are the opportunities? And I
13 think a lot of the discussions today will focus on what
14 some of these opportunities are. So how do we create
15 infrastructure around a development program for an
16 individualized therapy for which there may be no
17 similar products developed or licensed in the past? An
18 example of this is, again, individualized bacteriophage
19 therapy where there's no existing model or structure
20 for clinical development or a specific guidance for
21 this type of product. And as experts in the field
www.transcriptionetc.com
205
1 emerge and collaborate, we can build systems that
2 facilitate the development of multiple research
3 programs. And we hope to address that today.
4 The roadmap of this infrastructure will
5 necessarily include novel clinical development programs
6 that may include innovative clinical trial designs and
7 statistical approaches to small population-based
8 clinical studies. Ideally, these early approaches to
9 building infrastructure can create a foundational basis
10 for future development. If we're able to leverage both
11 prior and collective experiences to guide us forward,
12 we can expedite and optimize the development of these
13 products that can meet specific individual needs.
14 So in order to maximize these opportunities
15 that we identify to meet the challenges that we see, we
16 will require flexibility. And where are there
17 opportunities for flexibility? Today, we hope to
18 discuss some innovative approaches to the clinical
19 portion of our product development program, including
20 areas where regulatory flexibility may be applicable.
21 Some examples of this include, as I mentioned, novel
www.transcriptionetc.com
206
1 clinical endpoints and statistical approaches that
2 allow for the enrollment of small populations of
3 subjects or subjects that may receive different
4 products within the same clinical study.
5 So some of the major areas that we hope to
6 cover today in our presentations and our panel
7 discussion include: the approach to designing and
8 interpreting efficacy assessments in studies that may
9 occur for a single individual or in a small group of
10 subjects. Again, the example of this is the
11 development of infrastructure around approaches to
12 demonstrate clinical benefits with phage therapy.
13 Along similar lines, accumulating safety data across a
14 range of disorders for treatments with genetically
15 modified hematopoietic stem cells can be informative.
16 Our presenters will also be addressing the
17 need for study designs that facilitate the
18 interpretation of clinical data that may present unique
19 challenges, such as the use of bacteriophages
20 adjunctive therapy in the context of complex and
21 varying antibiotic regimens across a variety of
www.transcriptionetc.com
207
1 anatomic locations and etiologic agents. Of course,
2 another important consideration, and many people have
3 already addressed this in the morning session, are the
4 timelines for development of much needed individualized
5 therapies. Today, we'll discuss the impact of these
6 timelines on end-to-end development and approaches to
7 optimizing this timeline in the context of advanced
8 therapies.
9 As we move from our presentations into our
10 panel discussion, we hope to touch on the some of the
11 following points, and we're looking forward to a
12 thoughtful and insightful discussion with our patient -
13 - with our panelists, speakers, and audience members.
14 Some of these include the approaches that may
15 facilitate the assessment of efficacy in
16 individualized therapeutics, especially in the context
17 of small groups of subjects. Are there ways to
18 leverage accumulated safety data to enhance assessments
19 of safety? How do we approach the interpretation of
20 safety data in clinical development programs that may
21 not progress according to the usual paradigm? And what
www.transcriptionetc.com
208
1 are some statistical and study design approaches that
2 we can consider as we exercise flexibility?
3 So I'd like to introduce our first speaker.
4 This is Dr. Schooley, who currently serves as Professor
5 of Medicine and Senior Director of International
6 Initiative at UCSD, one of my alma maters. He
7 completed medical school and an internal medicine
8 residency at Johns Hopkins and an ID fellowship at NIH
9 and Mass General Hospital. He was at Harvard in 1981
10 with early research efforts directed at the
11 pathogenesis and therapy of herpes group and retroviral
12 infections. He was head of the Division of Infectious
13 Diseases at University of Colorado in 1990, another one
14 of my alma maters. And he led the NIH AIDS Clinical
15 Trials Group from 1995 to 2002. And he's currently, as
16 I said, now at UCSD, and he's serving as the head of
17 UCSD's Infectious Disease Division. And his recent
18 interests have focused on the use of bacteriophage to
19 treat infections. And we are excited to hear his talk
20 today.
21
www.transcriptionetc.com
209
1 OPPORTUNITIES AND CHALLENGES IN THE CLINICAL
2 DEVELOPMENT OF BACTERIOPHAGE THERAPEUTICS -
3 DR. ROBERT T. SCHOOLEY
4
5 DR. SCHOOLEY: Thanks very much. It's a
6 pleasure to be -- to follow someone who's had such an
7 illustrious pathway in terms of institutions she's been
8 in and to try to talk to you today a bit about a
9 juncture in phage therapeutics. I'm going to try to
10 bridge some of the elegant discussion you heard this
11 morning from Jason Gill about some of the aspects of
12 phage production, phage biology, say a little bit about
13 the -- where phage therapeutics are today and then to
14 go on to reach forward to the final discussion this
15 afternoon where we'll be talking about access to
16 emerging therapeutics with a few comments about how
17 those might be able to fit into some of the clinical
18 trial development approaches as well.
19 So, the -- today, I'm going to try to briefly
20 talk about some of the limitations of antibiot- --
21 anti-microbial therapy. Everyone here is quite aware
www.transcriptionetc.com
210
1 of most of that, say a little bit about how phage
2 therapeutics might fit into some of these gaps. I'll
3 say a small amount about what we know today, go on to
4 talk about some of the key gaps, and then talk about
5 moving from where we are today to orderly clinical
6 development.
7 Obviously, these days as you read the
8 increasing number of reports of multidrug resistant
9 bacterial infections, it's clear that the microbial
10 evolution on a global platform is outpacing our ability
11 to keep up with them with small molecule traditional
12 antibiotics. So one of the obvious places for phage
13 therapeutics is to deal with this because they've been
14 innovating in terms of any microbial activity for about
15 300 million years, and we only started about 80 years
16 ago. The -- we also know that phages do some things
17 that antibiotics don't do, for example, interfering
18 with biofilms. We know that phages can go to some
19 places that are difficult to penetrate with antibiotics
20 and places where antibiotics don't work as well. So
21 they're -- in addition to just spectrum, there's some
www.transcriptionetc.com
211
1 other aspects of phage therapeutics that are attractive
2 in thinking about how we might take advantage of these
3 activities in individualized applications.
4 So what are the opportunities here? There's a
5 virtually unlimited number of and diversity of phages
6 in nature. And we have tools now to genetically modify
7 those phages. Whether we will need to do that and what
8 we can do to make them better than they already are, I
9 think, remains an area of intense interest and one that
10 will have to be validated in the clinic in a series of
11 iterative clinical studies.
12 We know that there's some opportunities with
13 phages that are quite attractive, particularly the
14 pharmacodynamics. With antibiotics, we give a dose of
15 antibiotic, and we're already worried about whether the
16 kidneys or the liver or both are leaching the
17 antibiotic away from the site of infection before the
18 next dose. With phages, in theory at least, once the
19 site of infection is seeded, self-replication can
20 continue to provide a phage to deal with the
21 antimicrobial challenge.
www.transcriptionetc.com
212
1 We also know that phages can disrupt biofilms.
2 There have been an increasing number of animal studies
3 and clinical anecdotes in which this seems to be
4 playing a role. Phages have a theoretical advantage of
5 being less disruptive to the microbiome and
6 contributing to further antibiotic resistance when
7 they're used in -- as targeted therapeutics. And
8 finally, there are some reports of how phages can re-
9 sensitize organisms to antibiotics through both
10 mechanism-based approaches and more generally.
11 Now, where would you think about using phages?
12 Obviously, for patients that have organisms that are
13 not susceptible or to antibiotics that patients can
14 tolerate, places where we're not -- where antibiotic
15 delivery or activity is limited by the anatomy, and in
16 situations in which having something to deal with
17 biofilms might be particularly attractive. And these
18 are areas that are already under investigation and also
19 targets of individual experiences in the clinic.
20 So what do we know today? Well, we know this
21 morning that -- from Dr. Gill's comments, we can make
www.transcriptionetc.com
213
1 relatively large batches of phages quite safely and
2 quite uniformly. And we can administer them to the
3 patients with very little evidence of toxicity if the
4 phages are produced in ways that are cognizant of
5 contaminants and deal with issues related to genes you
6 might not want to have in the phage and so forth. So
7 the technology is there to produce phages in a very
8 homogenous way. There are a fair number of nuances
9 about phage stability in different conditions and host
10 strains and so forth that are quite different from the
11 antibiotic situation. But they're all solvable in this
12 context of individualized therapy.
13 We know that there have been a lot of -- a
14 large number of anecdotal cases in which phages seem to
15 have shown benefit for individual patients. But these
16 are all anecdotes. And we need to move from anecdotes
17 to more organized data to be able to understand how to
18 actually use phages in clinical practice. And we know
19 that, as with antibiotics, resistance develops quickly.
20 And we need to understand how to circumvent this with
21 phage cocktails and other approaches that are also
www.transcriptionetc.com
214
1 under development.
2 So what are the gaps? Well, one of the major
3 gaps is how do we know which phage or phage combination
4 is best to use in a given patient? If one remembers
5 back to the early antibiotic days, early antimicrobial
6 testing involved agar plates, broth dilutions. And we
7 spent a long time, 20 or 30 years optimizing predictive
8 mod- -- approaches for clinical efficacy, and we’re
9 still not there. Phages are the same story.
10 We have very little knowledge about what the
11 optimal dosing levels, routes of administration, and
12 duration of therapy should be with phage therapeutics,
13 the classical Phase 2 sorts of data that one would want
14 to have of an antibiotic before proceeding to clinical
15 endpoint trials. We don't know enough about the
16 antimicrobial activity of phages in vivo in humans. We
17 know that they can select for resistance, which is a
18 very good measure of demonstrating they have any
19 bacterial activity. But they -- how the antimicrobial
20 activity compares, for example, to antibiotics head-to-
21 head is something that is still up in the air.
www.transcriptionetc.com
215
1 We know that resistance develops quickly, but
2 we don't understand the determinants and the kinetics
3 of resistance and how that may be affected by things
4 like phage host range and other aspects of phage
5 biology. And we need to understand how to mitigate
6 this. And finally, we need to move on to think about
7 how to demonstrate phage efficacy in specific clinical
8 situations.
9 Now, where is phage therapy these days? Well,
10 there's been a lot of off-line use for over 100 years
11 in many parts of the world. And I think the take home
12 message for most of this is that, by and large when
13 phages are given orally or topically, there's very
14 little evidence of toxicity. But it's been very
15 difficult to assess objectively whether there's any
16 evidence of efficacy because many of the endpoints are
17 chosen post-hoc.
18 Many of the patient populations tested were
19 not homogenous, and very few of the phage preparations
20 used were well-characterized. We know that there are
21 increasing number of individual uses under eINDs. And
www.transcriptionetc.com
216
1 again, we're beginning to see that some of these are
2 becoming more organized and more standardized, in some
3 ways bridging the way to early clinical trials from the
4 standpoint of a little more homo- -- homogeneity.
5 So how might this eIND experience bridge
6 towards IND experiences? Well, one approach is that as
7 people who are confronting patients along with the --
8 often with similar clinical indications confer with
9 each other, we often will converge on the same
10 approaches and based on what other people have done so
11 that each patient isn't a brand new patient. I'll talk
12 to you a little bit about how the iPATH center at UCSD,
13 which has begun to both do clinical trials and assist
14 with clinical cases, has been trying to do that.
15 We also know that with -- standardized
16 clinical approaches do emerge when people do the same
17 thing over and over again. There was a very nice case
18 series in the Nature Microbiology that appeared in
19 paper this month and online last month about a dozen or
20 so patients with staph sepsis from Australia. And the
21 M. abscessus experience with Graham Hatfull, since
www.transcriptionetc.com
217
1 Graham is about the only one who has M. abscessus,
2 we've been collaborating with him. And most of the
3 therapeutic interventions there have been pretty
4 standardized because we really don't have many data
5 points to start variation.
6 Now, how have we tried to do some of this at
7 iPATH in terms of trying to, while the clinical trials
8 are up and going, make phages available to people who
9 need them? We have been engaged in this -- initially
10 stumbling into it about four years ago with a faculty
11 member at UCSD. And then one of our astute clinicians
12 began to identify other patients who needed phage
13 therapy. And since that time, another eight or nine
14 patients have been treated at UCSD. And then, we began
15 to get calls from other places about how to approach
16 phage therapeutics. And the Chancellor of UCSD put
17 together some seed money for us to put an
18 infrastructure together to be able to respond to some
19 of these requests as they came in.
20 The approach we currently take is, when we get
21 a request from a patient or family or treating
www.transcriptionetc.com
218
1 physician, is to try to get back to the treating
2 physician because we are not trying to be the patient's
3 doctor. And many times in those settings, things get
4 lost in translation, and we don't want to be giving
5 medical advice to people who are not in front of us.
6 After we talk to the doctor, often the doctor
7 will tell us what this patient needs is a dose of
8 ampicillin, or what this patient needs is nothing. And
9 we move on, or sometimes the physician will say, “You
10 know, I don't really know what else to do but would
11 phage therapy be reasonable?” If it seems reasonable
12 we'll talk through some of the options. And if the
13 physician wants to proceed, then we proceed to try to
14 help them do that, with the first step being trying to
15 help them find a phage product that can be used.
16 There aren't many sources for those these days
17 in terms of phage production and discovery operations
18 that provide phages of sufficient quality that we would
19 feel comfortable giving to patients in any kind of an
20 organized way. We have kind of -- there are a group of
21 collaborating institutions and investigators who have
www.transcriptionetc.com
219
1 expertise with specific organisms. People at Texas A&M
2 are particularly good at Burkholderia. Baylor works on
3 E. coli and Klebsiella. Yale has been very interested
4 in Pseudomonas and so forth.
5 And so what we'll often do is say to a
6 physician, “We don't have any solution for you, but you
7 might want to talk to Jason Gill or to someone at one
8 of the other places.” They will then send an isolate
9 to that location, and they will then -- at this
10 location, if they have agreed to try to screen, will
11 either come up with a phage or not. So one of the
12 bottlenecks is whether or not the receiving laboratory
13 can come up with a phage that -- or phage, preferably a
14 cluster of phages that might be active against the
15 patient's isolate.
16 If they find a suitable phage, there are a
17 couple of -- the next bottleneck is moving from an
18 academic laboratory discovery operation to being able
19 to have a product that be given to a patient
20 perennially. And many academic labs haven't done that
21 before. They have been -- they have very high-quality
www.transcriptionetc.com
220
1 operations. But things -- thinking about pharmacologic
2 stability, how to dispense the product over a period of
3 days or weeks from an investigational pharmacy, dealing
4 with things like bacterial contamination with a USP 71
5 testing and so forth, these are not things that many
6 academic labs think about. And one of the things we've
7 tried to do is help them at least become aware of
8 those.
9 And increasingly, these labs have -- are now
10 incorporating that so that when they talk to the FDA in
11 conjunction with the referring physician -- because
12 that's where the eIND has to come from -- many of these
13 things have already been taken care of before the FDA
14 reviewer has to say, “So what have you done about
15 sterility?” If they ask us about dosing we say, “We
16 don't know the answer to that, but this is what we've
17 done in the past. And these are things you might
18 consider.”
19 And what often happens is people will use
20 doses that are similar, routes that are similar to what
21 we've used before. And while clinical trials are
www.transcriptionetc.com
221
1 proceeding and being developed, we hope that some of
2 this homogeneity will help, at least within the agency,
3 begin to look at cases in more of a -- in the aggregate
4 than as each case being a different approach with
5 different doses given by different routes for different
6 durations. And the phage therapy that is given to the
7 patient is obviously given -- provided by the referring
8 -- by the laboratory and given under eIND.
9 So another thing that is happening under this
10 is, as time has gone on and more of these experiences
11 have evolved, there are certain very frequent flyers
12 that come up, including treatment of non-tuberculous
13 mycobacterial infections. Graham Hatful has been
14 collaborating with us for about two-and-a-half years
15 now and has had about 90 requests for treatment, mainly
16 of M. abscessus, mainly in trans- -- renal transplant,
17 liver transplant patients, immunocompromised patients.
18 And the bottleneck there has been mainly finding
19 isolates that are -- phage that are active against any
20 given isolate.
21 Having said that, because the patients are
www.transcriptionetc.com
222
1 similar, the requests are similar, we've developed a
2 shell protocol that we're currently trying to refine
3 and work with people at the NIAID to begin to think
4 about how this might move on into a real clinical trial
5 but, in the meantime, suggesting approach like this
6 when patient -- when physicians want to take this to
7 the FDA as eINDs before this emerges. Now, where do
8 you go from there? Well, it's not rocket science about
9 how to develop antibiotics, and these are antibiotics.
10 I think one of the things that has happened
11 with phage therapy is that it has been treated as if
12 it's something else. These are antibiotics; they just
13 happen to be living. And we've had paradigms to
14 develop antibiotics for 80 years. And if we don't use
15 those paradigms, taking into account some of the
16 biological variabilities, differences with phages, we
17 won't know how to use them in clinical practice.
18 And we won't know how to benchmark them with
19 antibiotics when choices are made. And we won't know
20 how to use them together because phages will not be
21 used instead of antibiotics except in some
www.transcriptionetc.com
223
1 circumstances. At least initially, they will be used
2 with antibiotics. We need to use them in a context
3 that we understand. And we developed -- we,
4 collectively as a community, have developed approaches
5 to evaluate antimicrobials.
6 Now, there are some nuances to phage
7 therapeutics that are different, and you have to think
8 about in terms of the typical Phase 1, Phase 2, Phase 3
9 design and issues related to CMC. We've -- some of
10 which were talked about this morning by Jason Gill. So
11 the clinical trials are -- have begun and one of the
12 challenges has been trying to think about how they can
13 be done in an orderly way.
14 You've seen most recently what amounted to a
15 Phase 3 trial done in Europe on burn patients before
16 Phase 2 data had been generated in terms of
17 understanding dosing, understanding stability on the
18 way to the patient, understanding phage interactions,
19 and even understanding the microbiology of what was
20 being treated. So we've had a lot of enthusiastic
21 efforts to get ahead of the curve without going through
www.transcriptionetc.com
224
1 the steps between here and there to develop the kind of
2 data to have studies be done in a more organized way
3 that you would do if you had an antibiotic.
4 Now, Phase 1 trials, if you start going
5 through the typical approach with phage therapeutics, I
6 would argue are almost useless at this point. We know
7 phages are basically safe if they are prepared well.
8 It doesn’t make a lot of sense to try to understand
9 pharmacokinetics and pharmacodynamics of phages in
10 patients that don't have an organism the phage will
11 grow in. So the traditional, “But have you given that
12 to a human yet,” whenever a new phage comes along is a
13 waste of time to talk about and really, I would argue,
14 don't help much anymore.
15 Moving to Phase 2, the major caveat I would
16 make about this is in -- with antibiotics we would
17 demand Phase 2 data before going to Phase 3. And if
18 any generalization can be made about drug development,
19 going to Phase 3 before you understand Phase 2 has been
20 a graveyard for drugs of all classes. And I think
21 phage therapeutics is one that needs to be careful not
www.transcriptionetc.com
225
1 to fall into that trap.
2 So one of the Phase 2 trials that is being
3 developed is one that is being supported by NIAID and
4 being carried out by the Antibiotic Resistance
5 Leadership Group. It's a very simple study trying to
6 understand the activity of phages in humans outside the
7 context of when antibiotics are given in conjunction.
8 This is a study that would be done in patients with
9 cystic fibrosis, who are clinically stable, and don't
10 need antibiotics at the time but are shedding
11 Pseudomonas aeruginosa chronically.
12 The inclusion criteria would be mainly -- we
13 would be looking at people who repetitively shed
14 pseudomonas in their sputum. There are quite a few
15 people like this who are clinically stable. The goal
16 here would be to do a standard single ascending dose
17 study to understand when phages are given by an
18 intravenous or by an aerosolized route, how long they
19 reside in the lung, what they're antimicrobial activity
20 is, what the evolution of the phage and the organism is
21 under treatment when used together, and to use that to
www.transcriptionetc.com
226
1 build a database to move on to do multi-dose studies
2 and studies along with antibiotics in Phase 3 trials.
3 It's a very similar -- very simple design of multiple
4 cohorts with a placebo in each cohort and, at the end
5 of the day, expanding the cohort that looks optimal to
6 get more precision around the measurement.
7 Now, Phase 3 trials are moving along. I've
8 already made the point that we need to, I think, be
9 careful about launching Phase 3 trials unless we
10 understand what we're doing because the worst thing for
11 this field would be several failed Phase 3 trials that
12 are done in a way in which the data were
13 uninterpretable. The -- it didn't work. See, they've
14 been trying for 100 years, here's another failure is
15 probably the biggest, I think, short term danger to
16 this field because it may or may not have promise. But
17 it would be a shame to have it put aside without
18 understanding the science under the hood.
19 So in terms of Phase 3 trials, I've already
20 made this point. These studies should be done as if
21 these are antibiotics. That's what they are. And we
www.transcriptionetc.com
227
1 need to have microbial endpoints. We need to put them
2 together with clinical endpoints. We need to think
3 about them in the kinds of trial designs that one would
4 use for treatment of multi-drug resistant organisms, if
5 that's what you're after.
6 We can think about other clinical
7 applications: implanted prosthetic devices in which you
8 might be trying to show, in a placebo-controlled way,
9 that you can salvage devices that would otherwise have
10 to be removed. There are many clinical trial designs,
11 but they should be thought about in ways that give
12 crisp endpoints and that are -- give you information
13 that will help you clinically.
14 In these Phase 3 trials, the unique aspects
15 that are different from antibiotics are that they --
16 the organism that the antibiotic in question actually
17 replicates after you give it at the site of infection.
18 And I think mathematical modeling will be very
19 important was we move ahead to understand relationships
20 between population sizes and clinical scenarios. We're
21 going to be in a situation where we will have to think
www.transcriptionetc.com
228
1 about aggregating patients with specific clinical
2 conditions, like E. coli UTIs, in clinical trials in
3 which each patient will be treated with a different
4 agent. And by using endpoints that are well-defined
5 and agents that are well-defined in terms of how they
6 are made, we should be able to develop systematic
7 information that will inform clinical use.
8 Finally, people have talked about phage
9 therapeutics as being high risk, high gain. I would
10 argue that's not really the case. They really are
11 potentially high gain, but are they really high risk?
12 Phage have been around for 300 million years, and
13 they're still here killing antibiotics -- or killing
14 antibiotics -- not yet. They're still here killing
15 bacteria. And what we've had over the last 15 or 20
16 years is a lot of advancement in terms of how to
17 prepare phage, how to work with them in laboratories,
18 how to purify them, and get them to the bedside.
19 And what we need to do now is to develop
20 rigorous clinical trials to figure out how best to use
21 them and to have us understand that the reason we do
www.transcriptionetc.com
229
1 clinical trials isn't to get drugs approved; it's to
2 learn how to use them in clinical practice. And so of
3 course the optimal clinical trial is when you can do
4 both at the same time, and I think phage therapeutics
5 is ripe for that at this point. Thank you very much.
6 DR. REINDEL: So our next presentation is
7 going to be conducted remotely due to travel issues.
8 Dr. Kohn is a professor in the Departments of
9 Microbiology, Immunology, and Molecular Genetics as
10 well as Pediatric Hematology/Oncology at UCLA. He's
11 board certified with more than 30 years of experience
12 in treating children in the clinical bone marrow
13 transplantation space.
14 His principle area of research is the
15 development and application of methods for gene therapy
16 of blood cell diseases using autologous hematopoietic
17 stem cells. His lab has investigated methods for
18 optimal gene delivery and expression and gene editing
19 with human hematopoietic stem cells performed in
20 clinical trials of gene therapy for genetic diseases
21 and pediatric HIV and AIDS. He's won many awards and
www.transcriptionetc.com
230
1 been appointed to many prestigious positions, and we're
2 excited to hear his talk today.
3
4 CHALLENGES TO DEVELOPING INDIVIDUALIZED STEM CELL GENE
5 THERAPIES - DR. DONALD B. KOHN
6
7 DR. KOHN: Well, thank you. Can you hear me?
8 Okay. Thank you. Talking to a phone, it's hard to
9 know. Sorry, my travel issue isn't a virus. It was
10 multiple mechanical problems on United Airlines that
11 couldn't get me there.
12 So I'm going to talk about work that we've
13 done over the last couple of decades actually to
14 develop gene therapy for a rare disease, ADA SCIDs that
15 I'll talk about, kind of, to show that, maybe, if
16 anything, this is the old school traditional route and,
17 at the end, talk about ways that we can possibly bring
18 these kinds of therapies about more quickly forward.
19 So the next slide is my disclosure. And it's relevant.
20 I'm on the board for a company called Orchard
21 Therapeutics, and my university has licensed IP to
www.transcriptionetc.com
231
1 them. And I'll talk about that today. That's the ADA
2 SCID gene therapy. Next slide.
3 So what I'm going to talk about is just give a
4 little background on sort of this area of therapeutics
5 targeting hematopoietic stem cells with gene therapy to
6 treat blood cell diseases. Then I'll talk about it as
7 a case example gene therapy for ADA SCID. And then
8 I'll close with some not so deep insights on lessons
9 that are learned that might be useful for development
10 of individualized therapies. Next slide.
11 So this is required by our union to be shown
12 at all presentations. This is a hematopoietic tree,
13 making the point that it's the hematopoietic stem cell
14 that lives normally in our bone marrow that both self-
15 renews and gives rise to all the blood cells. Next
16 slide.
17 And so hematopoietic stem cell transplants can
18 cure a whole list of genetic diseases of blood cells,
19 and this is now over 40, 50 years of work. We can take
20 allogeneic stem cells from a well-matched donor, or
21 even a haplo-identical donor now, and transplant them
www.transcriptionetc.com
232
1 to essentially replace the patient's own stem cells,
2 burying the monogenic defect to treat classes of
3 primary immune deficiencies. And I'll talk about SCID,
4 hemoglobinopathies like sickle cell and thalassemia, a
5 whole list of lysosomal storage and metabolic and
6 leukodystrophies, and congenital cytopenias.
7 So they -- it won't -- you can't treat
8 everything with hematopoietic stem cells, but there's
9 at least several dozen blood cell related diseases that
10 are macrophage, monocyte-related disorders that can be
11 treated by replacing the defective stem cells. Next
12 slide. And so, again, showing on this tree, so then
13 the technical task is that the gene correction event
14 needs to occur in the multi-potent long-term
15 hematopoietic stem cells. Everything after that is
16 sort of entrenched and amplifying effect cell, and the
17 effect would be short-lived if we put the gene into,
18 for example, a progenitor.
19 And so the two sort of major approaches that
20 have developed now are adding a gene using an
21 integrating virus, as I'll show you. Or the really
www.transcriptionetc.com
233
1 exciting emerging area that Keith touched on a little
2 bit in the therapeutic approach is to edit the gene in
3 the stem cell. And either of these if you make a
4 permanent change in the genome of the stem cell that
5 will then be propagated to all the blood cells that
6 follow. Next slide.
7 And so this is cartoon, sort of, the process.
8 And it starts with the patient. And their
9 hematopoietic stem cells are isolated, which is one of
10 the things that makes this therapy much easier than all
11 the other in vivo approaches that we take the cells out
12 of the body, then in the laboratory either add the gene
13 with a integrating vector and several of the types of
14 viruses that are listed there, or use the new -- all
15 the gene correction methods to site-specifically
16 correct the defect or knock out a gene or change a base
17 or whatever you want to do in the stem cells ex vivo.
18 And then, typically before the cells are given
19 back to the patient, the patient will receive some
20 types of chemotherapy, or in the future hopefully,
21 monoclonal antibodies to get rid of some of their own
www.transcriptionetc.com
234
1 hematopoietic stem cells to make space so that when you
2 give back this modest amount of cells that you've take
3 from the patient they can reconstitute a lot of their
4 blood cell production. Next slide. And so the field
5 has gone through sort of two major rounds of viral
6 vectors. And so the first generation of vectors shown
7 at the top were from -- typically from murine Moloney
8 leukemia virus, where the virus' long terminal repeats
9 were intact and had strong enhancer promotors that make
10 a lot of the transgene messenger RNA and transgene
11 protein then. But these were dangerous because they
12 had strong enhancers in their LTRs that can
13 transactivate an adjacent gene so that, when you add
14 these cells to 10 to the 8th cells from the patient's
15 bone marrow, they land relatively randomly. And if
16 they happen to land next to a proto-oncogene, the
17 enhancers could turn them on. And that in fact
18 occurred in some of the clinical trials in the 2000s.
19 So the field has largely turned to sort of the
20 types of vector shown at the bottom, these second-
21 generation self-inactivating, or SIN vectors, where the
www.transcriptionetc.com
235
1 enhancers are deleted from the long terminal repeats.
2 And then the gene can -- transgene can run off an
3 internal promoter that can be selected, either that
4 it's a reasonably strong promoter without a lot of
5 enhancer activity or, in fact, can be lineage specific
6 like beta globin to make beta globin for sickle- --
7 thalassemia or sickle cell. Next Slide.
8 And so using this approach -- and this lists a
9 number of the disorders now that have been treated, not
10 approved drugs yet but at least in preliminary Phase 1
11 and 2 trials, show evidence of clinical efficacy and
12 good safety. So many of the diseases that were on that
13 initial list have now been approached. And one of the
14 challenges here that, you know, is even more extreme in
15 individualized therapies is that each genotype requires
16 a separate vector carrying the gene and a developmental
17 project.
18 And so it sort of limits the development and
19 the number of disorders that are approached. But in
20 fact, the safety record from these vectors to the
21 present time has been quite good. There have been no
www.transcriptionetc.com
236
1 vector related serious adverse events that I'm aware
2 of. And one of the worries has always been, since
3 you're using a vector based on HIV, the potential for
4 elements of the vector and the packaging that come
5 together to make replication confident lentiviral
6 vectors that could spread. And to my knowledge again,
7 this has never been reported in either products or in
8 patients.
9 And using the lentiviral vectors, the
10 integration sites don't show preferential integration
11 near oncogenes. And there have not been any clinically
12 significant clonal expansions, again, that I'm aware
13 of. So they're looking relatively safe, although still
14 it's probably maybe 500 people or maybe 1,000 worldwide
15 that have received these kinds of vectors into
16 hematopoietic stem cells. So it's still relatively
17 early in the developing a safety base. Next slide.
18 And so then I want to talk about what's been
19 my favorite disease to treat for many years now, Severe
20 Combined Immune Deficiency or SCID. And SCID is the
21 most severe of the human primary immune deficiencies,
www.transcriptionetc.com
237
1 of which there's several hundred. SCID has absent T
2 and B cells, and NK function is variable depending on
3 the genetic type. And in fact, SCID can be caused by
4 more defects than any -- in any one of more than 20
5 genes. And in total, SCID is quite rare. About 1 in
6 58,000 is one of the best estimates from -- now that
7 there's newborn screening we're getting a much better
8 feel for the frequency. And SCID as a severe immune
9 deficiency has been uniformly fatal in infancy before
10 treatments were developed.
11 Typically, there'd be severe recurrent
12 infections, chronic diarrhea, failure to thrive leading
13 to death in infancy. And there was one famous child
14 who was maintained in a germ-free bubble for more than
15 a decade. And that's why it's sometimes called bubble
16 baby disease.
17 And we know that bone marrow transplant can be
18 curative. So we know that giving normal hematopoietic
19 stem cells can essentially replace the immune
20 deficiency. And in fact, if there is an HLA-matched
21 sibling donor, which occurs in about 20 percent of
www.transcriptionetc.com
238
1 patients, there's more than a 95 percent success rate.
2 And the small failures are typically patients who have
3 severe infections at the time of transplant. But for
4 the majority of patients that don't have a matched
5 sibling, the results have been less good. Although,
6 they continue to improve using either matched unrelated
7 donors or haplo-identical typically parental donors.
8 Next slide.
9 And so then, just the specific disorder that
10 we focused on, ADA SCID, it's the cause of about 10 to
11 15 percent of human SCID. So it's the second or third
12 most common gene that can cause SCID. And we estimate
13 there's about 10 children born a year in the U.S. and
14 Canada based on the referrals that we've had and the
15 population incidences. And ADA SCID patients have
16 profound pan-lymphopenia. So they have typically --
17 essentially no T, B, or NK cells at shortly after birth
18 from accumulating the toxic adenine metabolites that
19 ADA would normally be part of catabolizing.
20 So ADA SCID has been the focus because it was
21 the first genetic form of these more than 20 of human
www.transcriptionetc.com
239
1 SCID with a biochemical. And then the genes were
2 cloned sort of in the mid-'80s. And so it's the first
3 where things began because the gene was in hand. And
4 so as I referred to there, there are multiple
5 therapeutic options for patients including all the
6 allogeneic stem cell transplants from matched siblings,
7 matched unrelated, or haplo-identical donors.
8 There's also an FDA approved, and in fact two
9 serially, a purified one and now a recombinant, enzyme
10 replacement of polyethylene glycol modified ADA that
11 can be used to lower systemic ADA levels. And then
12 there's also emerging autologous stem cell transplant
13 gene therapy that I'll talk about. So the next slide
14 shows the lentiviral vector that we've worked with now
15 for the last seven, eight years.
16 So we've been doing trials sort of
17 successively over 20 years using the earlier type of
18 vectors that I showed you. And then about eight years
19 ago, talking to colleagues at the University College
20 London, Adrian Thrasher and Bobby Gaspar, we said we
21 should move to a lentiviral vector, and this one was
www.transcriptionetc.com
240
1 developed. So it has the human ADA cDNA running off
2 the elongation factor alpha core promoter, which is one
3 of those promoters I mentioned. It's a pretty strong
4 promoter but doesn't have much transactivating activity
5 of trans enhanced nearby genes.
6 And this turned out to be a very well-behaved
7 vector. So it's one of the SIN types of vector with an
8 elongation factor promoter. The cDNA is codon
9 optimized to get better expression. The WPRE elements
10 stabilizes the method so you get more bang for the
11 buck. You get more protein per transcript, and it's
12 pseudotype VSV-G. And it has a very high titer. This
13 is from our lab. We can get very high titers after we
14 concentrate it, so it goes into stem cells very
15 efficiently. Next slide.
16 And so we spent about two years once we had
17 this vector chosen doing the pre-clinical work that
18 comprised the IND pharmtox package. And so we looked
19 at efficacy in term of the transfer and expression of
20 the ADA gene in patient derived bone marrow cells, both
21 in vitro, and then put into immune deficient mice in
www.transcriptionetc.com
241
1 vivo. And compared to a retroviral vector that we were
2 using before, MND ADA, the EFS ADA lenti had higher
3 gene transfer and higher ADA production per vector copy
4 number. So we had good activity.
5 And then we did a series of safety studies.
6 We put this vector into bone marrow of either ADA
7 deficient mice or to human cells in immune deficient
8 mice. And in fact, either the retro or the lenti,
9 neither showed any leukemia or clonal expansion in
10 these models, maybe suggesting these models are not
11 very robust. Integration set analysis did show that
12 the retroviral vector was more often near the
13 transcriptional start sites in cancer related genes
14 that the lentiviral vectors.
15 That's kind of a very well-known recurrent
16 pattern of integration for these two classes of
17 vectors. And in fact, in an in vitro assay, the murine
18 gamma retroviral vector caused murine lineage negative
19 bone marrow cells to clonally expand using an assay
20 called in vitro mutagenesis assay and the lentiviral
21 vector didn't. And so in fact, we had used the gamma
www.transcriptionetc.com
242
1 retroviral vector that we're comparing here for several
2 trials beforehand and have had no adverse events from
3 the vector. But these pre-clinical studies showed the
4 lenti was at least as safe and possibly safer. And so
5 that's sort of those studies which we published
6 composed the toxicology and pharm package for the IND.
7 Next slide.
8 And so I just wanted to take -- you know,
9 what did it take to do? So this is the slow road to
10 lenti, which is a pun. And so this shows you in the
11 academics what it took us to get from, “Hey, we should
12 make a lenti” to an open trial. And so the gree- --
13 highlighted in green are the funding applications where
14 we got funding in various stages to do the -- pre-
15 clinical work was done on a program project grant we
16 had from heart, lung, and blood. Heart, lung, and
17 blood also had a gene therapy resource program that
18 paid for the GMP comparable vector for the pharm-tox
19 study. And then the NIAID, we received an U01 award
20 for the clinical trial, for the Phase 1 trial.
21 And we went through a regulatory gauntlet of
www.transcriptionetc.com
243
1 the RAC, the FDA pre-IND. This study was initially
2 opened up for both UCLA and the NIH. So we had IRBs
3 and IBCs at both places. And then, so we submitted the
4 IND at the end of 2012, so about four years from sort
5 of proof of concept to an IND. And I think probably
6 that can be done more quickly. We were sort of
7 learning as we were going. Next slide.
8 And so in fact, we opened up this Phase 1
9 trial in the U.S., and a parallel trial opened up in
10 London where we treated patients with this vector in
11 low dose relatively Busulfan conditioning. And we'd
12 roll patients who had ADA SCID without a matched
13 sibling donor. They had adequate organ function and
14 could not have an ongoing active infection.
15 And our primary endpoint was safety, survival,
16 event free survival, event being, sort of, failure and
17 needing to have a rescue transplant, and scored adverse
18 events. And then secondary endpoints were more for
19 efficacy, measuring the production of gene marked blood
20 cells from the stem cells we treated, looking at immune
21 reconstitution, and clinical endpoints of infections,
www.transcriptionetc.com
244
1 and hospitalization. So the trial opened up in May
2 2013 under an IND. Next slide.
3 And we actually went through and treated
4 patients relatively quickly. The initial plan was to
5 treat 10 patients, but we kept getting referrals. So
6 we wound up treating 20 patients. So over the course
7 of about five years, we went from the first patient
8 into the last patient visit treating 20 of the ADA SCID
9 patients. Next slide.
10 And these are some early interim data. These
11 aren't the official data, but these are data when
12 about, I think, 15 of the patients had been -- were out
13 at least a year, looking at various outcome parameters.
14 And so, red blood cell ADA went from zero because of
15 their ADA deficiency to, in fact, slightly above the
16 normal range for red cells from healthy donors. Their
17 bad metabolites, the deoxyadenosine metabolites dropped
18 down. When these patients initially present untreated,
19 these dAXPs are in the 50 percent or higher. And then
20 you can see that they also had immune reconstitutions,
21 so their T cell numbers and B cell numbers came up
www.transcriptionetc.com
245
1 after the gene therapy, after we stopped their enzyme
2 therapy. Next slide.
3 And so the second part of the slow road to
4 lenti -- so I sort of showed you the first half of
5 this. And so at the bottom half, so I just talked
6 about the Phase 1 and 2 trial. In fact, we decided we
7 should move to producing the cells in a cryopreserved
8 formulation and planned that as we then -- as this
9 property was licensed from our university, UCLA, to
10 this company, Orchard Therapeutics. So all this -- and
11 I moved into a new world of commercialization. And
12 they've then taken the ball from that point, and I will
13 be submitting a BLA application for licensure. Next
14 slide.
15 And so then, this slide just shows sort of for
16 this disease kind of the timeline from discovery of the
17 cause to treatments. And so ADA SCID was observed sort
18 of serendipitously being present in a few babies with
19 ADA SCID by Eloise Giblett back in 1972. Then we moved
20 forward to identifying the gene, cloning it, making
21 vectors and then a series of clinical trials. And in
www.transcriptionetc.com
246
1 fact, a gamma retroviral vector for this disease is
2 approved in the European Union as a drug called
3 Strimvelis which is available for therapy. And then
4 there's the lentiviral vector. So next slide.
5 So I showed this list before. I'll skip this
6 slide. Next one. So that's the work it took us to
7 develop a treatment for one of the hundreds of PIDs.
8 And in fact, I'm aware of currently that there are
9 three genotypes of SCID in gene therapy trials: ADA
10 SCID, X-linked SCID, and Artemis SCID. One form of
11 chronic granulomatous disease, Wiskott-Aldrich Syndrome
12 and Leukocyte Adhesion Deficiency, I believe those are
13 all the immune deficiencies currently being treated by
14 gene therapy, but in fact there's many others. So
15 there's another at least 17 other genotypes of SCID,
16 four other genotypes of CGD. And there's a whole list
17 of other even rarer monogenic primary immune
18 deficiencies that could be treated by this approach.
19 So the question is what would be needed to
20 develop individualized therapies for these other even
21 rarer primary immune deficiencies using an analogous
www.transcriptionetc.com
247
1 approach? Obviously, it's, you know, we can't spend 10
2 years, and companies are not going to invest in
3 diseases that are going to be treating, you know, two,
4 three patients a year. And then beyond the immune
5 deficiencies, there's all the other blood cell diseases
6 that would fall under this treatment: red blood cells,
7 white cells, platelets, stem cells, and then, you know,
8 even beyond that, other genetic diseases. Next slide.
9 So I guess the question is, you know, how do
10 we do this in less than 45 years? I mean, that's
11 obviously intolerable for patients who have diseases
12 that need to be treated right away or very soon. And
13 so over this period that I just showed you, the
14 development of this treatment, the investigative
15 capacity of biomedicine has vastly expanded as we've
16 all witnessed.
17 So we have far greater resources now,
18 reagents, materials, all the multiple -omics and
19 informatics to really move things quicker. And in
20 fact, we now have established a number of broad gene
21 manipulation capabilities both virally, vector addition
www.transcriptionetc.com
248
1 of AAV or lentis, and all the editing approaches that
2 Keith talked about briefly with growing positive
3 experience for safety and efficacy.
4 I think, you know, when this started out when
5 I -- at the beginning of the field, there was, you
6 know, it was unknown what kind of problems might
7 develop. And I think we are -- although there have
8 been some problems certainly along the way, we are
9 developing a growing experience of safety and efficacy.
10 And so we now have this cumulative experience in gene
11 and cell therapy product development, pre-clinical
12 evaluation, manufacturing, and clinical trial
13 performance. Next slide.
14 So how would newly identified genetic
15 disorders be fast tracked for individualized therapies?
16 Well, as someone said earlier this morning, the first
17 step is to understand the pathogenesis so we can
18 understand the therapeutic approach. One issue, you
19 know, is it an absent gene product that we just need to
20 add back the gene, or is there an abnormal dominant
21 negative or dominant adverse gene product that needs to
www.transcriptionetc.com
249
1 be overridden or knocked out?
2 And then we need to understand, you know, what
3 are the relevant cell targets? So if it affects the
4 blood cells, then what I was talking about would be do
5 allogeneic stem cell transplant or ex vivo gene therapy
6 may be beneficial. Other disorders like CNS, the
7 defective microglia, that are -- many of which are
8 blood cell derived can benefit. But this won't help
9 all the other organs most likely.
10 If it affects -- is an autosomal recessive
11 disease affecting motor neurons, then IV or intrathecal
12 routes might be needed. If it's an autosomal dominant
13 disorder affecting neurons, we may need to deliver the
14 cells or genes in situ. Deficiency of serum proteins
15 made in the liver, then intravenous AAV looks like a
16 very viable approach to treat those. And so there
17 won't be one size fit approach for all these genetic
18 diseases. It's really going to depend on which organs
19 are involved and what's the nature of the defect.
20 So we have a number of models to sort of work
21 this up in, and one of the most important always is
www.transcriptionetc.com
250
1 patient derived cells that have the defect that can be
2 studied. And now, we have great capabilities to make
3 induced pluripotent stem cells and make organoids that
4 can recreate elements of the disease. But, of course,
5 all the murine knockout and gene manipulation models
6 provide a way to test the new therapeutics.
7 And then, you know, once we've developed it,
8 we need to define the nature of it and, you know, is it
9 something that we've experience with? So is it just
10 AAV for a new genotype of a retinal disorder, for
11 example? Is it a cell type we've used, or is it
12 something new, and how much experience to do have also
13 then with the cell type hematopoietic stem cells, T
14 cells, liver cells, et cetera? Next slide.
15 So what are some of our opportunities to try
16 and use this experience for other diseases? Well, one
17 is that we can rapidly identify these responsible rare
18 gene defects underlying the inherited and de novo
19 diseases, whole exon, whole genome, CGH, et cetera. We
20 can relatively rapidly develop vectors and CRISPR-based
21 therapeutic targeting reagents. You know, within a few
www.transcriptionetc.com
251
1 weeks to months, we can have reagents targeted to a
2 specific disorder.
3 And as I said, the record certainly for lenti
4 and AAV vectors are mature with expanding safety
5 records. So it raises the question of how much pre-
6 clinical testing is needed if you just change a
7 transgene or you just change a small guide RNA. There
8 are clearly potential adverse events from a different
9 transgene or a different guide, but the more we can
10 leverage platforms and experience, the quicker it will
11 be. And then we can use gene engineered murine human
12 iPSC as I mentioned in patient derived cells to
13 determine the disease modifying activity. Next slide.
14 So some of the challenges still though are
15 quite significant. And one of them that has plagued
16 gene therapy and cell therapy since the beginning is
17 the challenge of in vivo delivery. So in vivo
18 delivering cells, genes or editing reagents to specific
19 cell types, and sites and tissues remains really
20 suboptimal in many cases. And I think that's still
21 limiting and, you know, I think is a very important
www.transcriptionetc.com
252
1 area. We're all hoping that there will nano techno- -
2 - nanoparticle technologies that will do that, but
3 that's still an early area I'd say.
4 And then beyond sort of the simple monogenic
5 disorders that I've talked about is there are many
6 other more complex genetic disorders that are
7 chromosomal deletions or duplications. Those are going
8 to be much more challenging to treat by either
9 replacing the large deletions or selectively removing
10 duplicate segments than are these single gene targets.
11 And then many of our diseases obviously are multigenic,
12 and these would be much more complex to approach by
13 either gene addition or editing methods. And these
14 cases, cell therapies that have the whole package might
15 be better.
16 And so I think Dr. Marks in his opening
17 remarks talked about a four to eight year typical
18 timeline, and I think I would agree with that. And so
19 the breadth of activity in toxicological testing for
20 pre-clinical studies that are typically done to support
21 IND are expensive and take a long time. And I was
www.transcriptionetc.com
253
1 probably low-balling at a $500,000 to $20 million and
2 one to three years. And I will stop there and look
3 forward to the discussion. Thank you.
4
5 PANEL SESSION WITH Q&A
6
7 DR. REINDEL: I'd like to invite our panelists
8 up to the front here, please. So in addition to Dr.
9 Kohn, who will continue to participate in the panel
10 discussion over the phone, we also have several FDA
11 representatives, including Dr. Lapteva from the
12 Division of Clinical Evaluation Pharmacology and
13 Toxicology and the Office of Tissues and Advanced
14 Therapies and Dr. Xu, who is a Senior Mathematical
15 Statistician in the Office of Biostatistics and
16 Epidemiology.
17 So I see no one has approached the microphones
18 yet. I'd like to encourage the audience to do so. If
19 you have questions, we'd really like to engage in
20 discussions, but I'll be happy to start the discussion
21 off with a question that I think is relevant to both
www.transcriptionetc.com
254
1 Dr. Kohn and Dr. Schooley. I was really impressed with
2 the way that both of you sort of united resources
3 across the country to enable collaboration in a space
4 that -- where you may be able to only enroll one or two
5 patients with a certain condition. Can either or both
6 of you talk a little bit about strategies that have
7 been effective to promote that kind of collaboration?
8 DR. SCHOOLEY: I think the main thing is
9 really being open to collaboration and realizing we're
10 all facing the same problems there. We gain by
11 collaborating. And there has been a lot of publicity
12 about MDR infections, so patients drive a lot of this
13 as well. And I think listening to patients and
14 physicians and their needs and trying to meet them is
15 one of the things we should do as investigators. I'm
16 sure that -- let's move to a genetic perspective on
17 that as well.
18 DR. KOHN: Yeah. And if I can comment, so
19 we've done a few trials for ADA SCID and CGD, for
20 example, with two or three or four different academic
21 sites. And besides having the colleagues to
www.transcriptionetc.com
255
1 collaborate with, which is enjoyable, spreading it out
2 geographically is helpful for patients so they don't
3 all have to go to one place in the country. And one of
4 the things that we've learned that's very important,
5 obviously, is having very careful monitoring of the
6 sites so that everyone is doing things exactly the
7 same.
8 And obviously, drug companies do this all the
9 time when they're doing multi-centered trials. But
10 coming from the academic perspective, it's a lesson we
11 had to learn to have ongoing active monitoring of the
12 clinical data, and also, in our trials at least, we've
13 mainly been doing cell manufacturing at each academic's
14 GMP site. And to harmonize that activity takes a lot
15 of work because, in general, people think they all know
16 how to do it. And -- but if you work with sites that
17 are agreeable, you can sort of standardize even
18 something as relatively complex as processing
19 hematopoietic stem cells with viral vectors.
20 MS. MCGRATH: Yeah. Lynne McGrath. I was
21 curious about your comment that you're hoping that it
www.transcriptionetc.com
256
1 doesn't take 45 years. And I'd like to turn the
2 question on to the group to say that was a technology
3 that has evolved over the last 45 years, and I've
4 personally been involved with programs that took over
5 20 years. But the question is, when new technology
6 emerges today, how do we not let that go 45 years?
7 Because certainly we have a lot of things that
8 have happened. But just your thoughts on some of the
9 new and emerging scientific discoveries that -- how do
10 we get that baseline information to be able to use
11 those as therapies? You know, the 45 years may still
12 be -- hopefully not but may still be something that we
13 would consider because of, you know, going forward with
14 new discoveries. So that's kind of a question that I
15 have is how do we shorten that? I don't know if
16 anybody has any thoughts.
17 DR. KOHN: Well, I mean, one example are the
18 use of CRISPR. So you know, that was only really
19 identified, what, seven or eight years ago or
20 something. And they're already -- it's already in
21 clinical trials. And so clearly, we have accelerated
www.transcriptionetc.com
257
1 our process of -- from discovery to clinical
2 applications.
3 It'll still need to go through now the process
4 of Phase 1 and Phase 2 trials to some extent. But I
5 think it's such a nimble platform for developing a
6 therapeutic if you're targeting a gene that I, you know
7 -- I think that will be one thing that will, for
8 genetic diseases at least, really accelerate the
9 timeline.
10 DR. SCHOOLEY: You could argue that phage
11 therapeutics has been going on for 100 years, so that
12 45 years is nothing. And I think the real key is
13 trying to understand what you're doing in as precise a
14 way as you can so you learn from it and can generalize,
15 and other people can either repeat or improve on the
16 experience. Where we make mistakes is where we do
17 things without characterizing them as carefully as we
18 can to be able to learn when it doesn't work, to learn
19 from things when they do work, and to build on what we
20 know.
21 So I think that should be probably the most
www.transcriptionetc.com
258
1 critical aspect of trying to accelerate discovery
2 across all fields. And I think it fits what was said
3 earlier today by some of the other people talking about
4 platforms that different companies might be developing
5 for different applications and indications. Why not
6 learn from each other? You're not competing for the
7 same product, and the more you share the more you can
8 focus on things that matter which is your particular
9 application.
10 So I think that phage therapeutics are the
11 same way. Companies aren't competing with each other.
12 It's not like people are lining up to decide which
13 company to invest in. People are trying to decide
14 whether this is crazy or not, and so the more
15 collaboration the better.
16 MS. WITKOWSKY: Hi, this is Lea Witkowsky from
17 IGI at UCSF and UC Berkeley. Hello, Dr. Kohn online.
18 DR. KOHN: Pleasure.
19 MS. WITKOWSKY: I'm wondering, all this talk
20 about the ability to separate product specific
21 attributes and processes from platforms seems like a
www.transcriptionetc.com
259
1 really important distinction, if we can come up with
2 ways to streamline creating evidence and being able to
3 leverage that evidence from one application to the
4 next. And since we're mentioning genome editing, I'm
5 wondering, Dr. Kohn in particular, as you've worked in
6 -- with hematopoietic stem cells across various
7 different diseases, how much do you expect an organ
8 system -- so for example, platforms of using lentivirus
9 in one organ system or tissue system to be
10 standardizable so that, if you're doing genome editing
11 for example, you're simply changing one component, kind
12 of leveraging the modularity aspect of things like
13 genome editing? I wonder if you could speak to that as
14 you've worked on multiple different diseases within the
15 blood organ. Thank you.
16 DR. KOHN: Yeah. No. Thank -- that's a
17 really good question. You know, I don't know fully the
18 answer. So if we use the example of using CRISPR-Cas9
19 to modify specific sites, obviously every time you go
20 to a new guide, it has a different on-target, off-
21 target profile. But as to -- by what Keith talked
www.transcriptionetc.com
260
1 about, our development to really identify that, you
2 know, that may be some -- you know, that kind of safety
3 analysis may be enough to let you not have to do the
4 full developmental package, just to target a different
5 genomic site, for example.
6 And the same thing with lentiviral vectors.
7 When you look at all the different papers now that have
8 been published looking at integration sites for
9 lentiviral vectors with a number of different diseases,
10 it's getting very monotonous because you kind of see
11 the exact same pattern. So we know that, and so it
12 then just becomes what are the transgene specific
13 issues that need to be studied. Obviously, you need to
14 show disease activity modification.
15 But beyond that can we, based on the class of
16 gene -- that it's a metabolic enzyme for example -- not
17 have to do all the extensive testings I showed you that
18 we did in mouse models, transplant models, et cetera.
19 And so I think some of it -- I think some of that is a
20 regulatory issue of what will be acceptable to allow
21 existing data to be leveraged and not have to start at
www.transcriptionetc.com
261
1 square one every time you change a guide or a cDNA.
2 DR. REINDEL: We have some questions from the
3 online audience.
4 THE OPERATOR: So we have two questions from
5 online. They're both pretty short, so I'll give them
6 to you at once. The first one is Dr. Schooley
7 mentioned about how phages are similar to antibiotics.
8 How do we ensure the environmental safety of phages?
9 Do we have any knowledge of how they might affect the
10 safety of the microbiome outside the patient? That's
11 question number one. And question number two is how
12 can the phage and gene therapy fields learn from each
13 other?
14 DR. SCHOOLEY: Well, question number one is
15 we're all loaded with phages, and we have -- we walk
16 around with more phages than we give to patients.
17 They've been around for 300 million years, and when
18 there is not substrate for them to grow in, they no
19 longer propagate. So phages are kind of self-renewing
20 and self-extinguishing when their substrate is gone.
21 One of the things that make them, I think,
www.transcriptionetc.com
262
1 less dangerous environmentally than antibiotics is
2 their spectrum is so narrow. So although I certainly
3 understand the issue, unless we were to come up with
4 some genetically engineered phage that had broad host
5 range, would take out all but one particular bacterium,
6 and that were, for example, Pasteurella pestis, I think
7 competition among phages will take care of that. And
8 the other question was related to --
9 THE OPERATOR: To field learning from --
10 DR. SCHOOLEY: Oh, I'm sorry. Learning from
11 each other. So let's, Dr. Kohn, see if there's
12 anything to learn and vice versa. Sorry.
13 DR. KOHN: Well, yes. I mean, I think, again,
14 I tend to think that a lot of these types of novel
15 therapies emerge from academic medical centers. And we
16 all spent our time in medical school and not doing
17 manufacturing. So I think some of the CMC issues that
18 you touched on we've also needed to learn and develop.
19 So I think, although we're making a different product,
20 we're using similar processes. So I think this bit of
21 crosstalk is useful. And maybe this is a Gordon
www.transcriptionetc.com
263
1 Conference topic or something to have adjacent meetings
2 and some overlap and talk about developing these kinds
3 of therapeutics.
4 DR. SCHOOLEY: Some of the same issues in
5 terms of producing phage are ones that Dr. Kohn is
6 talking about because different academic labs are
7 producing phages of different types but are beginning
8 to converge in terms of how they're purified. And
9 those kinds of convergences, I think, are very helpful
10 and make it easier to think about what's being done in
11 different places as well.
12 DR. KOHN: Right. And standardization of
13 potency testing and even titering is something that, at
14 least for lentiviral vectors, is totally lab specific
15 what a titer value is because there is not standardized
16 method. So I think, again, we face a lot of the
17 similar issues in these products even though they're
18 different products.
19 DR. REINDEL: If there are no additional
20 comments from the audience, I had another question that
21 I think will be helpful to the audience to hear. Could
www.transcriptionetc.com
264
1 you, Dr. Schooley or Dr. Kohn, discuss any specific
2 challenges you've faced in the design of these clinical
3 trials in terms of the challenges that you discussed
4 just now in terms of product and the approaches you've
5 taken to overcome those challenges?
6 DR. SCHOOLEY: Well, I think the challenges
7 that are common to both are -- have to do with the fact
8 that we're -- that the interventions are quite
9 individualized from patient to patient. And we have to
10 think about how to, as one person put this morning,
11 talk about the process by which they are made and what
12 standards and what metrics are used to say that these
13 products are similar enough that, when you use them in
14 different patients directed at the same organism but
15 with a different, for example, host range, that you're
16 -- you can aggregate the data in a generalizable way.
17 So being able to characterize the products in
18 a way that you can talk about their potency, talk about
19 their host range, what receptors they use, things that
20 let you, again, know what you're giving makes it easier
21 to design a hypothesis generated trial. That looks at
www.transcriptionetc.com
265
1 both clinical endpoints and biology and to -- at the
2 same time you're looking at the clinical endpoints,
3 learn enough about whether you're delivering the --
4 just like you're delivering a gene to the cell you want
5 to get to, we have to deliver the phages to the site of
6 infection, know they stay there, know that they remain
7 active, and measure those things at the same time the
8 clinical trials are being designed.
9 So what the challenge is there is that many of
10 the companies that are developing phages are, as
11 therapeutic agents, are relatively thinly capitalized.
12 And it's very difficult for them to support a lot of
13 the translational research that needs to be done. So
14 finding ways to get that done at the same time, I
15 think, is critical to moving the field as a whole
16 forward. But I'll stop there because there are
17 obviously other things that have to do with oncology
18 that are important, too.
19 DR. KOHN: Right. Well, so of course number
20 one is always funding. So as I showed on the timeline,
21 we had a -- every six months to a year or so we had to
www.transcriptionetc.com
266
1 apply for another set of funding to make it to the next
2 step. So I think CERM in California has done a very
3 nice job of sort of laying out their funding mechanisms
4 to follow the developmental timeline. And I think some
5 of the NIH institutes are also moving towards sort of
6 more multi-stage funding, so you don't need to back to
7 complete new R01 application for each stage of the
8 product -- project. So that's one of the issues.
9 The timelines are long for developing these.
10 And for academic careers it's not the best thing if
11 it's going to take you, you know, 10 years from when
12 you start to when you have your Phase 1 trial done.
13 It's hard to become an associate professor if you start
14 as an assistant with that. And so that's a challenge.
15 And then, you know, it's -- they're expensive.
16 To manufacture these products at high quality costs a
17 lot of money for the GMP, for the testing, the staff,
18 and also then the clinical trials are expensive. So I
19 guess that comes back to my first point of funding.
20 And so, you know, I think those have been the
21 challenges. We've obviously overcome them because
www.transcriptionetc.com
267
1 we're talking today, but those, I think, probably limit
2 what's developed.
3 DR. SCHOOLEY: Stability of funding is really
4 critical in phage therapeutics as well. The -- we've
5 seen over the last -- this trial that we talked about
6 briefly today has been in the works for two-and-a-half
7 years and watched multiple companies come and go, each
8 of which has gone down because of inability to maintain
9 their development plan. So it's critical to have
10 overarching mechanisms and support approaches that let
11 you plan for something that be carried out from
12 beginning to end. And having the government support
13 some of the basic and translational work, really, I
14 think move both fields forward in a way that get us
15 products, which is what we're all trying to do.
16 Because that, at the end of the day, is why we go to
17 the lab.
18 DR. KOHN: Yeah. And just to comment -- and
19 these challenges are probably even more intense for
20 individualized therapies, to get back to the theme,
21 that, you know, a one-off therapy that might be
www.transcriptionetc.com
268
1 lifesaving, if it costs a lot, if it takes a long time,
2 won't be valued. And so we have to find ways to do
3 these quicker and cheaper. Not sure what that answer
4 is, but I think that's the challenge.
5 DR. LAPTEVA: Yeah. So I would like to make a
6 comment and perhaps address some of the questions that
7 were asked earlier about product specific versus
8 platform and how we approach individualized
9 therapeutics and how if we have only one patient that
10 would need to be treated with a particular therapy, how
11 can we make this clinical development program looking
12 efficient and really deliver to the patient who needs
13 the therapy? A number of people this morning and the
14 afternoon spoke about the need to digress, to some
15 extent, or maybe apply regulatory flexibility to the
16 traditional medical product development model in order
17 to make the development of individualized therapeutics
18 more efficient. And although at this stage we don't
19 know collectively how these development programs may
20 look like and it's likely there would no one size that
21 would fit all and some of them will be very different
www.transcriptionetc.com
269
1 from others, we could at least try to identify some the
2 factors that would influence this digression from the
3 traditional development.
4 So one potential factor that we've heard about
5 today, and I could foresee, is the ability to make a
6 reasonable prediction about the product effect at the
7 time when the decision to treat for therapeutic
8 purposes is made. Another is the general expectation
9 that individualized therapeutics should work in
10 patients for whom they've been designed. And one other
11 important aspect, I think, is the determination of the
12 dosing. This is one of the very challenging aspects in
13 the development of individualized therapeutics.
14 But speaking about the decision to treat, if
15 you look at the traditional model of product
16 development, the decision to treat the disease for
17 therapeutic purpose typically does not come into the
18 picture until later because when people participate in
19 clinical trials, and when it is a group setting where
20 some people are treated with the investigational
21 therapy -- and the drug is the same for everybody --
www.transcriptionetc.com
270
1 and some people may be treated with placebo, and some
2 may be treated with active comparator --it depends on
3 the clinical trial design -- there is a little more
4 acceptance from the perspective of the patient as well
5 as the investigator that the product may not work. It
6 didn't work. It was investigational. This was a
7 clinical trial which failed to demonstrate the product
8 effectiveness.
9 If the product does work and there are
10 appropriate statistical methodologies that support the
11 positive therapeutic effect of the product, then we can
12 make an inferential conclusion that the patient
13 population with the disease will likely be benefitting
14 from this product. So when the next patient with the
15 disease comes to their physician in clinical practice
16 and the decision to treat is being made, then both the
17 patient and the physician have already some information
18 that gives them the ability to reasonably predict the
19 treatment effect because you would know that patients
20 who had the same disease, maybe similar
21 characteristics, were treated in clinical trials. And
www.transcriptionetc.com
271
1 you have some ability to predict the magnitude of the
2 effect of the product, whether it would or would not
3 work, and to understand some basic toxicities.
4 With the development of individualized
5 therapeutics, the decision to treat comes much earlier.
6 Even at the investigational stage, there is this
7 expectation that the product should work because it was
8 designed for this -- for the particular patient. Yet
9 the step which is so common for all of the different
10 products about taking the representative sample of
11 patients, testing the hypothesis, observing the
12 effects, and then translating it to the population with
13 the disease is absent. So there has to be something
14 that fills the gap with individualized therapeutics.
15 It would serve the information -- as the information
16 with the sufficient predictive capacity to enable this
17 decision to treat.
18 And so speaking of the novel technologies,
19 what we see although in very, very few examples of such
20 individualized therapeutics developments -- and thank
21 you to the person who spoke this morning who is the
www.transcriptionetc.com
272
1 patient with cancer. You didn't realize probably, but
2 you were one of the examples of what we're actually
3 seeing. We're seeing development of new methods and
4 methodologies that are based specifically on that
5 individual patient's genetic parameters, physiological
6 parameters, understanding of the cellular metabolism,
7 how cellular phenotype may be changing with the
8 introduction of a transgene in that particular patient.
9 So what we will likely see in the field with the
10 development of individualized therapeutics is the
11 growth and development of these methodologies that are
12 predicting the individual patient's response to that
13 particular individual therapy. So that's one.
14 The other is the expectation that with these
15 types of treatments we will see positive treatment
16 effects, if not to say large treatment effects. So if
17 you take a gene therapy, for example, which is targeted
18 to correct a functional gene or if you take a cell
19 therapy that's intended to replace some lost functional
20 cellular tissue, then you would expect that not one and
21 not two, but many physiological processes, downstream
www.transcriptionetc.com
273
1 physiological processes will be affected. And likely
2 also maybe some anatomical changes will occur which
3 would result in previously unseen quantitative or
4 qualitative effects of the therapy previously unseen in
5 the disease progression.
6 And in that case, and my clinical colleagues
7 will understand me, something that we call minimally
8 clinically important difference will not be hinged
9 anymore on the comparison between the two groups. But
10 it would be very important to receive the input from
11 the patients and their caregivers. And that's where
12 the collaboration between the patient community and the
13 investigators would be very critical in understanding
14 as to what is the meaningful effect of that individual
15 therapy or a number of individual therapies and not
16 only that but also how these effects progress over time
17 and how we can monitor and evaluate the effects of
18 individual treatments in the long term.
19 And we have to create systems to do this. And
20 I'm sure the next panel will be talking about it --
21 systems that are able to collect clinical data. But
www.transcriptionetc.com
274
1 not only that, but to go back and maybe -- somebody
2 also mentioned it this morning. If you have a
3 predictive statistical model which feeds and includes
4 physiological variables and IT variables and other
5 variables, why can't you fit -- feed the clinical data
6 back into this model to make it a little more
7 predictive? And so this is something that we will also
8 likely see developing.
9 And lastly, for collection of safety data,
10 particularly when it is a platform based product where,
11 say, a vector treatment that's been optimized already
12 with understanding how different elements of the vector
13 may interact with one or more transgenes that would
14 potentially be inserted for treating different diseases
15 -- if you take this type of platform and we have a
16 clinical data collection attached to it and you may
17 call it a master protocol or a platform protocol -- but
18 it would be important to incorporate common variables
19 into that platform to enable perhaps a meta analytical
20 activities to be done later down the road specifically
21 with regard to safety because safety could potentially
www.transcriptionetc.com
275
1 be evaluated across different diseases for very related
2 but slightly changed products.
3 DR. REINDEL: Okay. I think that concludes
4 the panel discussion for today. Thank you for all of
5 our -- to all of our participants.
6 DR. KOHN: Thank you.
7 DR. RAYCHAUDHURI: So, I'd like to thank all
8 the speakers and panelists and the audience for sharing
9 your perspectives. And I'd like to thank Dr. Reindel
10 for moderating the session. And Dr. Kohn, I'm very
11 sorry about your travel challenges but so very happy
12 that you could give your presentation and join the
13 discussion also. So we're going to --
14 DR. KOHN: Thanks for having me, Gopa.
15 DR. RAYCHAUDHURI: Thanks so much. So we will
16 take a short 10-minute break. And we have three
17 excellent speakers coming up in session four in what I
18 know will be a very thoughtful and very thought-
19 provoking session. So please take a quick break and
20 join us in ten minutes. Thanks.
21 [BREAK]
www.transcriptionetc.com
276
1
2 SESSION 4: PRODUCTS TO PATIENTS
3 DR. RAYCHAUDHURI: Okay. If everybody could
4 take your seats, we'd like to get started with session
5 four. So I just want to say, in addition to everybody
6 in the room we've had -- we have over 400 people online
7 who are following the workshop. So it's great to see
8 this level of interest.
9 So it's my pleasure to introduce Dr. Celia
10 Witten. Dr. Witten is the Deputy Director of CBER, and
11 she will be moderating session four, which focuses on
12 how to get products to patients in a timely manner and
13 maintain access for patients in a sustainable way. Dr.
14 Witten.
15
16 SESSION 4 MODERATOR INTRODUCTION: DR. CELIA WITTEN
17
18 DR. WITTEN: Thank you. In the prior three
19 sessions, we talked about the scientific and clinical
www.transcriptionetc.com
277
1 aspects of development and heard about the importance
2 of regulatory flexibility leveraging knowledge across
3 applications, the challenges in manufacturing and
4 testing, and the need for collaborations. There will
5 be three speakers in this session: Jill Wood from
6 Phoenix Nest, Dr. Alison Bateman-House from New York
7 University Langone Health, and Dr. Phillip Brooks from
8 NIH. For the panel discussion, the speakers will be
9 joined by Captain Julie Vaillancourt from the Rare
10 Disease Program at CBER and Dr. Chip Schooley who
11 participated in the prior session.
12 The focus of this session is on ethical
13 issues, collaborations, and stakeholder roles in the
14 end-to-end development of individualized therapeutics.
15 In general, the role of individual stakeholders can be
16 quite different in the development of products for rare
17 diseases than in the development of products for common
18 disease indications. This may be even more so in the
19 development of individualized therapeutics intended for
20 one or a small number of individuals.
21 As stakeholders take on certain roles in the
www.transcriptionetc.com
278
1 development of these products, including close
2 collaboration with other stakeholders, certain ethical
3 issues arise. Furthermore, the focus of collaborations
4 may need to go beyond development to consider future
5 sustainability issues, since not all products or
6 approaches may be commercializable. In addition, new
7 development paradigms may pose ethical conundrums that
8 are not necessarily features of the standard drug
9 development paradigm. Therefore, these three issues
10 are somewhat related.
11 Some of the ethical issues that we see are
12 noted on this slide. One question that needs to be
13 addressed for any drug product and development is the
14 determination of sufficient manufacturing and safety
15 information for a trial to proceed and also, as we've
16 heard, sufficient information to develop other aspects
17 of the trial, such as determining the starting dose for
18 a product that can only dosed once. So these kinds of
19 questions about manufacturing safety information may be
20 even more of an issue for products for a single
21 individual or a small number of patients because
www.transcriptionetc.com
279
1 testing can be resource intensive.
2 Note that the determination of whether there's
3 sufficient information for a trial to begin is not a
4 simple risk-benefit question because there may be
5 considerable uncertainty in the safety testing or, for
6 that matter, an assessment of activity. Often the
7 discussion regarding N of 1 development is focused on
8 the acceptability of administering the product to the
9 individual. But the boundary between research and
10 clinical care may be very unclear. However, we need to
11 make sure we learn from each clinical investigation,
12 including single subjects.
13 Some of the development programs have been
14 funded or championed by a patient, family member, or a
15 small group of patient families. While this is
16 commendable and there have been some striking
17 successes, it involves a heroic effort on the part of
18 these family members, many of whom describe the effort
19 as being equal to more than a full-time job. Do we
20 need to figure out what to do to ensure that this is
21 not the expectation for families and caregivers? And
www.transcriptionetc.com
280
1 that is one of the issues regarding funding of
2 development. Another potential ethical issue,
3 particularly with regard to funding, is the question of
4 how development decisions are made, for example, which
5 patients get treated. Dr. Bateman-House will be
6 providing her perspective on ethical issues that arise
7 in this area.
8 Patients and patient groups have always been
9 stakeholders in medical product development, but these
10 groups are increasingly taking a lead role,
11 particularly for products with rare diseases. We will
12 hear about an outstanding example of these efforts from
13 Jill Wood in this session. Academic developers play a
14 larger role in the development of many of these
15 therapies than traditional pharmaceutical companies, in
16 many cases developing and performing early clinical
17 testing and, in some cases, partnering with
18 pharmaceutical companies or forming small companies to
19 shepherd the product across the finish line.
20 Philanthropic organizations are playing an increasingly
21 significant role in development programs for some
www.transcriptionetc.com
281
1 pharmaceuticals and will be important in this area
2 also.
3 We've already heard from a number of speakers
4 that collaborations are needed. These collaborations
5 are needed for many reasons, which have been discussed
6 in previous sessions. Some of the products may have
7 limited commercial viability. Academic developers may
8 have limited resources. In addition, as previous
9 speakers have noted, information sharing to eliminate
10 duplication of expensive development work would help
11 these products move forward.
12 There are many different models for
13 collaborations, and some amazing individuals and
14 organizations are currently leading collaborations to
15 develop individualized therapeutics. And there are
16 many models for information sharing. We can learn from
17 these examples as we work together to find a way to
18 develop and make available such products in a way that
19 is ethical and as resource efficient as possible.
20 However, it's possible some new models of
21 collaborations are needed because development work may
www.transcriptionetc.com
282
1 be duplicated, and these innovative areas benefit from
2 information sharing. There may be something to be
3 learned across patients for each product platform, as
4 we've heard during the prior sessions. And this could
5 include what we learn for a gene therapy vector for
6 related applications that could be related by tissue
7 target or by disease.
8 There could be a bio-distribution study done
9 for a vector that there'll be something gained in our
10 knowledge for applications with similar tissue targets
11 and delivery. And we may learn more about test methods
12 that will help in product development by gaining an
13 understanding of how these test methods work in
14 different applications. So it's important not just to
15 think about how to develop the therapy for each
16 individual or small group of patients but how we can --
17 you have to scroll down. Can you scroll down? -- how
18 we can do that sustainably across a range of
19 applications so valuable information is not lost or
20 overlooked but can be added to. Thus, we need to think
21 about development more holistically than one patient at
www.transcriptionetc.com
283
1 a time when we can. And we're going to hear a great
2 example of an effort in that direction from the talk
3 that P.J. Brooks will be giving during the session.
4 One last point I want to mention, we've been
5 referring to sustainability in this workshop. Why is
6 this important and why do we mention it when we discuss
7 collaborations? The questions of how to sustainably
8 provide products is important because, once a product
9 or a platform for designing individualized products has
10 been developed, how will patients be able to receive
11 these products? The traditional model has been for
12 pharmaceutical companies to take over production and
13 delivery. But some of these products may be valuable
14 to patients but not necessarily commercializeable.
15 Next slide.
16 So you've seen versions of this slide in each
17 of the previous talks illustrating the challenges and
18 opportunities in this -- in the area of product
19 development. This slide illustrates the fact that
20 traditional roles of how products are developed from
21 discovery through marketing are being upended, with
www.transcriptionetc.com
284
1 discovery being driven by patients and advocacy groups
2 as well as by NIH and other research grants funding or
3 by pharmaceutical companies. This upending of
4 traditional roles is seen all the way through
5 development. The question is where are the
6 opportunities for stakeholder collaboration among these
7 stakeholders and end-to-end development of
8 individualized therapeutics so we can move the field
9 forward?
10 Examples of areas for collaboration can
11 include availability of GMP grade material for clinical
12 studies. We have heard, both at this meeting and at
13 other venues held, how this challenge is limiting
14 development of AAV vector-based gene therapy. If this
15 same GMP grade material were available across multiple
16 researchers, for each product and development across
17 researchers, shared safety testing information could
18 help to reduce development costs for each product in
19 development for each researcher.
20 In addition, shared clinical data for
21 leveraging understanding would become possible.
www.transcriptionetc.com
285
1 Development of templates to facilitate IND submission
2 for collaborative development program might involve
3 multiple products but also could streamline
4 development. So as I mentioned, Dr. Brooks in this
5 session is going to describe NIH efforts of a
6 collaboration that's aimed at addressing these issues.
7 There are challenges for collaborations, also.
8 I've only listed a few examples on this slide, and I'm
9 sure the audience can come up with many more. But
10 including -- included among the examples would be
11 funding and governance. And by governance, I mean how
12 decisions are made regarding the collaboration goals
13 and the process, including the kinds of development
14 decisions I mentioned earlier when I discussed ethical
15 issues. And then there's the question of how
16 intellectual property is treated which also has come up
17 in this meeting.
18 And the last item, sustainability, I've
19 already mentioned, which is that we have to think about
20 not just development but what happens after the
21 development is completed because the goal of all of our
www.transcriptionetc.com
286
1 efforts should be to make sure our patients are able to
2 benefit from these new development programs. And
3 therefore, thoughts about how to do technology transfer
4 for commercialization are important, and also how to
5 ensure continuing availability.
6 I summarized at a high level the background
7 for the three related issues in this session: ethical
8 issues, collaborations, and stakeholder roles. And now
9 I'd like to introduce the first speaker. Do you have
10 the -- So our first speaker is Jill Wood. She's the
11 co-founder of Phoenix Nest, Inc., and she also has
12 funded a foundation to look at -- to try to develop
13 natural history studies and to try to develop funding
14 for Sanfilippo Syndrome. So please welcome Jill Wood.
15
16 THE TRIALS AND TRIBULATIONS OF DRIVING A TREATMENT FOR
17 AN UBER-RARE DISEASE TO THE CLINIC AND BEYOND-A
18 PARENT’S PERSPECTIVE - MS. JILL A. WOOD
19
20 MS. WOOD: Hi. Thanks for having me here. I
21 am Jill Wood. I am the mother of a child with an uber-
www.transcriptionetc.com
287
1 rare disease called Sanfilippo Syndrome Type C. And
2 I'm gonna walk you through our journey of what it's
3 been like trying to create a treatment and bring it to
4 the clinic for this disease. I really want to thank
5 the FDA for having me here today. It means a lot to
6 me. I greatly appreciate the fact that the FDA is
7 starting to look at the issues coming from the families
8 that are driving the science here.
9 So my clicker -- so before we can get to our
10 gene therapy utopia, we need a need. My need was born
11 on July 30, 2008. Jonah was full term. He was
12 absolutely perfect. We had no idea that he harbored
13 this insidious disease.
14 It was at our first year well visit that our
15 pediatrician noted that Jonah's head circumference was
16 off the charts, and we should probably go and get it
17 checked out. So we did. With due diligence, we went
18 to get our MRI, which was done at NYU, very fortunate
19 that we landed at NYU where the lab technicians there
20 knew exactly what they were looking at, which is
21 extremely rare in a disease like this. So they
www.transcriptionetc.com
288
1 suggested to our geneticist that we do a panel screen
2 for the MPSs.
3 So real quick about the diseases, I don't want
4 to bore everybody with any more science. Sanfilippo
5 Syndrome is MPS III. There are several different
6 MPSs. You've probably heard of many of them. The top
7 portion of those MPSs all have treatments. Sanfilippo
8 is the only version that does not have a treatment.
9 And I have four sub-types here: Type A, B, C,
10 and D. My son has type C, and my company is focused on
11 a treatment for type C and type D. And you can see
12 that our diseases are 1 in 1.5 million, which estimate,
13 maybe there's 100 kids in the United States by that
14 number. But I only know of 20 for type C and four for
15 type D.
16 Because our disease was so rare, we were told
17 that nobody was picking us up, that nobody would touch
18 us. One venture capitalist actually told me he
19 wouldn't touch my disease with a 10-foot pole. Yeah.
20 So you know, this was my first child. My husband and I
21 had done everything right. We had bought our first
www.transcriptionetc.com
289
1 house, paid off our school loans, and here you're gonna
2 tell me that my child has a terminal illness and
3 there's nothing I can do about it.
4 We hit the ground running, and we called
5 people that cared. We found these guys off of PubMed.
6 We brought our physicians -- fortunately, again, I'm
7 from Brooklyn, and I'm surrounded by wonderful
8 hospitals with geneticists and neurologists that were
9 ready to jump on our bandwagon and help us out. We had
10 a meeting in 2011, and we sat down with a few patients
11 that I had found, and our physicians and our
12 scientists. And we hammered out what it was that we
13 were going to do and that was to go for gene therapy.
14 So I'm going to go through -- the FDA asked me
15 how hard it was, you know, what did I have to do and
16 how we could do it better And so my first learning
17 curve was working with academia and finding the
18 scientists to help you start your program, getting the
19 -- your mouse model made. There's Alexi Bedeski with
20 our first mouse. I asked him to name him Juniper
21 because I wanted to name my second child Juniper. But
www.transcriptionetc.com
290
1 I'm not gonna ever have any more children, so there's a
2 Juniper.
3 But all of our science was funded outside of
4 the United States. It began in Montreal, with Brian
5 Bigger up there, and in Manchester. We funded these
6 guys through grass roots fundraising. We nickeled and
7 dimed it. That's a garage sale, a picture of a garage
8 sale there. This is how we did it.
9 It was a learning curve for me. We had to
10 write grants, hired lawyers, made sure that our
11 scientists were held accountable, that we had
12 milestones, and their payments were conditional on
13 their milestones. But for the people on the phone,
14 it's hard working with academia. These guys have a
15 school schedule. They take a lot of time off. And
16 sometimes they're post-docs, graduate, or they want to
17 get married or something. And you have to hire someone
18 else. It's not easy.
19 So during this process of nickel and diming
20 and trying to find -- scraping every dollar that you
21 possibly can, I met a gentleman who suggested that I
www.transcriptionetc.com
291
1 create my own company and go for NIH small business
2 grants. And I was like, you know, that sounds like a
3 great idea. Let's go for it. So we applied for an
4 STTR, these small technology transfer grants. I hope
5 many of you know who they are, what they are.
6 And unfortunately, there's small business
7 technology grants, and they don't want to fund
8 researchers that don't live in the United States.
9 Makes sense. So my type C research has not yet been
10 funded, but during this time I created a knockout mouse
11 for MPS III D. I actually applied for a competition,
12 Assay Depot, I want to give them some props for this Be
13 HEARD contests. I applied and we won the main prize, a
14 knockout mouse.
15 And that was 10 years ago about -- no, maybe
16 eight years ago, and that was like winning a car. You
17 know, it was before CRISPR, so that was pretty amazing.
18 Then, I went back, and I licensed our gene therapy
19 program from Manchester, brought it back into the
20 United States, and we're now working with our dosage
21 study with a CRO here. And I'm being helped with
www.transcriptionetc.com
292
1 additional funding from our friends at the Cure
2 Sanfilippo Foundation. Sorry.
3 So where we did not do so well with type C
4 grants, we have excelled with III D grants. And again,
5 major props to the NIH, we are now well on our way to
6 the clinic for MPS III D with an enzyme replacement
7 therapy. That one -- that started off from winning
8 that knockout mouse and snowballed into almost $7
9 million in SBIR grants. And like I said, there's four
10 patients, and those come from two different families, a
11 set of identical twins that you can see here, and a
12 younger family as well.
13 So while I'm developing treatments for type C
14 and D, I'm watching my sister diseases Sanfilippo type
15 A and B, which are much more prevalent than type C and
16 D. And it was really very exciting. They had several
17 programs in the pipeline. then all of a sudden, look,
18 Alexion/Synogeva, their ERT in MPS III B was shelved.
19 It was shut down; the trial was shut down before it
20 even got to the end. Same goes for ERT for -- with
21 Shire. BioMarin, if you might've heard from the press
www.transcriptionetc.com
293
1 releases, they did divest and were found a partner,
2 Levits. So that, thankfully, that trial will continue
3 on. But here we're sitting in limbo with Sobi, who
4 also wants to divest their ERT for MPS III A. You can
5 imagine how devastating this is for our community.
6 This is Will. Will's parents were told the
7 same thing that I was, that their child had an ultra-
8 rare disease and there was no treatment. A few months
9 later, lo and behold, here's a trial, and Will was
10 accepted into it. It says Shire right there. I'm
11 sorry to beat on Shire. That was actual Alexion's
12 trial. So sorry, Shire. That was actually Alexion
13 that dropped that trial, and they shelved it. They
14 dropped Will cold turkey. Took him straight off of his
15 ERT and is back to being told, “I'm sorry, your child
16 is going to die.”
17 Why is this happening? There's a lot of
18 speculations. I, for one, think that they have chosen
19 wrong endpoints. They're looking for cognitive changes
20 when our children have profound brain damage. And
21 we're not gonna change the cognitive in our children,
www.transcriptionetc.com
294
1 and it's really not what we're looking for. Financial
2 risks, a lot of these companies as you can see, have
3 changed hands, and not everybody's on the same page as
4 the previous CEO. BioMarin, bless them -- have nothing
5 against BioMarin -- but they went in a different
6 direction. They're not doing ultra-rare diseases
7 anymore.
8 And then, I think people don't realize our
9 diseases are slow to progress, but Will, Jonah, they
10 could die in their sleep tonight. It just happens.
11 Did I turn off my slide? Okay. Oh, but that -- okay.
12 This slide just kind of goes to shows you how
13 much our patient community has done. I am very, very
14 proud to be part of this Sanfilippo community. I don't
15 know that I've ever seen another rare disease community
16 work as hard as we have. Again, there's four different
17 sub-types, and we have four different enzymes. And
18 there's some things that are different about us.
19 So take for example these six programs that
20 have gene therapy programs in the works right now. The
21 first one nationwide was entirely -- pre-clinic was
www.transcriptionetc.com
295
1 entirely paid for by these foundations. And I think I
2 forgot to write the Cure Sanfilippo Foundation up
3 there, too. But all these foundations came in and
4 funded the pre-clinical, and then Abeona came in and
5 licensed it. Right now, they have A and B in the
6 clinic, and hopefully, it'll stay. And we'll have our
7 first treatment for Sanfilippo.
8 Lysogene was started by Karen Aiach.
9 Unfortunately, her daughter Amelia passed away just
10 before November and did not benefit from this
11 treatment. The trial still goes on underneath Sarepta.
12 I think I heard somebody from Sarepta here, so thank
13 you Sarepta for picking this trial up.
14 And then we have Estevee out in Spain, lay
15 low. Notable, Amicus has picked up MPS III A and B.
16 No, John Crowley is not a Sanfilippo parent, but he is
17 a Pompeii parent. Orchard was entirely funded pre-
18 clinical from the U.K. MPS Society. And, I think it
19 was the Ormond Street Hospital as well. And then
20 there's Phoenix Nest, who's driving the science for
21 type C and D.
www.transcriptionetc.com
296
1 The patient organizations, just, I mean, does
2 this shock the people in the audience that this has
3 been funded by us? And it shouldn't be this way. I
4 mean, we have children at home that are dying, and
5 we're here working our butts off trying to create
6 treatments for the next generation.
7 Okay. Don't you love the emojis on Apple? So
8 and this is also -- I had to throw this in there
9 because this really annoys me. It keeps me up at
10 night. Our SBIR grants, you know, you have a
11 collaboration with academia, and they get subawards.
12 They get a nice substantial subaward. They don't --
13 pays for all their overhead.
14 And, you know, they get to ask for some, a
15 bioreactor, you know, a $300,000 bioreactor. They get
16 some nice equipment. And then they get licensing
17 rights because it happens on their -- in their hallways
18 and on their property. And then they get all the fame
19 and glory and get to write papers and publish and go
20 around.
21 But then, I have to go back and license what
www.transcriptionetc.com
297
1 it is that my company won. I have to go and license
2 this. And what I don't think people really realize is
3 that I have to hire a lawyer. I have to hire patent
4 lawyers. Do you guys have any idea how much lawyers
5 cost? SBIR grants do not cover lawyer fees.
6 So to add insult to injury, here we are just
7 trying to scrape by on a disease for four kids. I
8 spent -- I actually spent over a year fighting this.
9 No milestone payments, no upfront fees. I mean, it's
10 just ridiculous to ask that from me. When you do, they
11 don't realize that you're sucking any incentives that I
12 had for commercial partners away. When you're treating
13 four kids, how many patients -- how much money are you
14 gonna make off of this?
15 So I'm throwing that out there. If there's
16 something we can -- some template we could put on our
17 SBIR grants that -- what do I want to say -- makes both
18 sides happy but realizes that these are ultra-rare
19 diseases and you can't treat me like a Parkinson's drug
20 company. Oh, boy. Things that I think the FDA, or the
21 NIH, could do for us that are no brainers, and I know
www.transcriptionetc.com
298
1 you know that they're no brainers because we've talked
2 about them a lot, natural history studies. One good
3 one, one we started -- we knew we needed to do a
4 natural history study.
5 But my families did not want to fund a natural
6 history study. They are extremely expensive. And when
7 you have $1 million dollars, do you want to send it
8 sending your kid to a clinic to be poked and prodded,
9 or do you want to put the money in making a mouse model
10 and making a gene therapy? I mean it's the mentality
11 of it. The families want to fund the research. And
12 now we're stuck.
13 We really, really have got to get our natural
14 history study undergoing. Mouse models, I have them.
15 But there are so many rare disease groups that don't
16 even have mouse models. And they're extremely easy to
17 make now and to house. I think if there was something
18 that the FDA could do for us, it would be to go through
19 and find out who doesn't have their mouse models.
20 Registries, registry's another major
21 contention amongst our patient groups. It's something
www.transcriptionetc.com
299
1 that we absolutely have to have as well, but again,
2 they extremely expensive to maintain. You have to be
3 HIPAA approved, GDPR approved, and those things have to
4 happen yearly. And who's going to do that? Who's
5 going to keep up on that? We really need help managing
6 and maintaining that. And that's, I think, another
7 thing our federal government could do for us and could
8 do it flawlessly.
9 And then one last topic I wanted to sneak in
10 here because nobody really likes to talk to their
11 patients about this -- patient groups. But biobanking
12 -- oh, there is a major lack of donations out there.
13 And if we could have this sensitive conversation with
14 our families that it is imperative that we keep some of
15 these tissues, brains, eyes, so you know -- it would be
16 extremely helpful. I think we talked a lot about
17 mutations as well and knowing the mutations of these
18 families. And I like this company, and the NIH
19 supports them as well. And I'm putting that up there
20 so people can take notes at home.
21 Here comes our Orphan Drug Act. And we love
www.transcriptionetc.com
300
1 Abby Meyer and what she did for us. That was done
2 almost 25 years ago, and I think it's about time that
3 we update our Orphan Drug Act. We all know that there
4 are 7,000 rare diseases, and here we consider that 1 in
5 20,000 with a patient population of 200,000. That's a
6 rare disease. So if you're a drug company and you're
7 gonna create a drug for a rare disease, who are you
8 gonna pick? One with 100,000 patient population or a
9 patient population of 20?
10 We're starting to talk about ultra-rare quite
11 a bit, but what is that number? What does that look
12 like? How do you go from 200,000 to 100? So if you're
13 gonna -- I think we need to get ultra-rare on the map.
14 We might as well get uber-rare on the map as well. And
15 we really need to figure out ways that we can
16 incentivize this.
17 I've been doing this for nine years. My drugs
18 are ready to go to the clinic. We're handing
19 everything over on a silver platter, and I still do not
20 have anybody knocking on my door. Pediatric review
21 vouchers are absolutely amazing, but it's still not
www.transcriptionetc.com
301
1 trickling down to us.
2 Just to let you know, we did do our pre-IND
3 meetings for both type D and type C. They went very
4 well. We went in really early because we were scared
5 about the slowly progressive heterogeneity, the tiny
6 patient population, and the lack of natural history.
7 How were we going to pull this off?
8 And CDER and CBER both gave us the same advice
9 was get creative and get more natural history, which we
10 are trying. I'm trying so hard to get our natural
11 history off the ground. And we will be looking at
12 mosaic endpoints, kids being their own control. We're
13 hoping to hang our hat on the guidance that was put out
14 last June and using heparan sulfate as our surrogate
15 marker.
16 Okay. You don't have to stare at this long.
17 You can look over at me, but you know, you read what
18 Sanfilippo looks like on paper. You read what these
19 diseases look like on paper, but to see it for yourself
20 is a huge difference. Yes, this is my bathroom at 8:00
21 in the morning. “Mom, I had an accident.” Okay, I'm
www.transcriptionetc.com
302
1 going to go and clean up his pants. You know, you're
2 parents. Somebody has an accident in his pants you
3 might throw them in the garbage. I look at my husband
4 and I --"Can we just throw the toilet in the garbage?
5 I can't deal with this.” Imagine this at Starbucks.
6 Picture that. Okay. You're never going back to that
7 Starbucks again.
8 Some families, their children scream and cry
9 nonstop for months on end, and nobody can figure out
10 what it is. Is it neurological? Does my child have a
11 bladder infection? Is it tooth decay? What is going
12 on here? And it al- -- it seems to end up being
13 neurological.
14 But these kids, I mean, it's terrifying for
15 the families. They don't sleep for months on end, and
16 these kids are mobile. So the families have to put in
17 safe rooms and lock the doors and lock everything down,
18 and their kids just walk around the room with the
19 lights on, switching it on and off, turning the T.V. on
20 and off. They're up all night long. It's horrible.
21 Imagine that.
www.transcriptionetc.com
303
1 So anyhow, yeah. This goes back to the point
2 that we don't care about cognitive endpoints. We need
3 to take care of some of the serious issues that we live
4 with day in and day out that will make our lives
5 better. I threw this in there, and we talked a little
6 bit about the workshop that happened last month for the
7 expanding AAV manufacturing capacity.
8 And that happened here, as well. And if -- I
9 encourage you guys to go back and watch this
10 conference. And I bring this up as -- for the end-to-
11 end gene therapy. And this company, I was very much
12 impressed with this company, the Discovery Lab, which
13 is doing exactly what we're -- we want to see happen.
14 I am now in the process of creating our
15 vector, putting it all together. This is another huge
16 hurdle for me and mind blowing. Three different
17 plasmids made at three different CROs that all have to
18 come together at one place and, if one isn't working,
19 then it throws everything off and you have to go back.
20 I mean, the level of expertise needed here is huge.
21 It's very time sensitive and the error for margin is
www.transcriptionetc.com
304
1 huge. So I commend this group for making this happen.
2 I would love to see it trickle down to the FDA for the
3 ultra-rare diseases as well.
4 Again, this was from that last conference, the
5 AAV conference. Last month alone I had three new
6 families from diseases that I had never even heard of
7 before calling me up that found me from random places
8 and say, “Hey, I need help. This is what's happening
9 to my child,” and it sounds very similar to what's
10 happening to my child. Each one of those diseases
11 breaks down into several other different subtypes.
12 They're all conducive to gene therapy. Just pull out
13 the gene and put in the next one.
14 And finding these gene therapy scientists, I
15 hate to even put pictures of Steve Gray up there. I
16 haven’t even talk about him because I don't want to
17 share him. But we have got to have more Steve Grays
18 out there and how it is that we're going to train these
19 people and, if they are there, to come out of the
20 closet somehow. You know, reach out to the FDA or the
21 NIH or Global Genes or EveryLife Foundation and say,
www.transcriptionetc.com
305
1 “Hey, I want to help. I have a lab; I'm interested in
2 gene therapy.” And these families are ready to
3 fundraise for you.
4 And yes, to my rescue, Jude Samulski, you
5 might have heard of him. His company AskBio has spun
6 out for Viralgene, and they have created a nonprofit
7 company where they are holding suites for uber-rare
8 diseases for people like myself. And we talked about
9 the bottleneck, these guys are holding a suite for our
10 gene therapy. So thank you very much.
11 Now, I always have to throw in my newborn
12 screening because I think this is vitally important for
13 all of our children, whether we have treatments or not.
14 Ignorance is bliss. That's my baby, a few weeks old.
15 But knowledge is power. Like I mentioned that Jonah
16 was diagnosed very early, he is the youngest child
17 known to ever be diagnosed asymptomatic without an
18 older brother or sister.
19 Because of that, he had tubes put in his ears.
20 He was a year-and-a-half. I didn't know he couldn't
21 hear. Put the tubes in, it gushed out, and he was
www.transcriptionetc.com
306
1 pointing out airplanes the next day. I was like he
2 couldn't hear that it was an airplane. It breaks your
3 heart to know that your child -- he can say ball, and
4 mom, and dad, but he couldn't hear. He also had bad
5 sight, so we got glasses.
6 His behavior, you can imagine during the
7 formative years how important it is to be able to hear.
8 Not only does it help you read and sit still and
9 participate in class, but it takes away that
10 frustration and helps with the behavior. Sanfilippo is
11 strife with behavioral issues. The kids are very --
12 can be very aggressive. So I actually feel guilty that
13 my son is doing better than any other Sanfilippo child
14 that I have ever met, and I attribute that single
15 handedly to the fact that he had early intervention and
16 we caught his hearing before he was deaf.
17 So that brings me to a close, and I have some
18 people that I want to thank. And lots of people aren't
19 on there. I can't thank everybody enough. There have
20 been plenty of professionals that have helped me with
21 pro bono services. But most importantly, I want to
www.transcriptionetc.com
307
1 thank my families.
2 We're very diverse. I have friends all around
3 the world. We don't speak each other's languages, but
4 we're family. And if it wasn’t for them, I'd probably
5 quit by now, but I know that they need me. And if it's
6 not for our kids, it's for the next generation because
7 no family should ever be told their child has a
8 terminal illness and there's nothing you can do about
9 it. Thank you.
10 DR. WITTEN: Thank you very much. Our next
11 speaker is Alison Bateman-House. She's an Assistant
12 Professor in the Division of Medical Ethics at NYU
13 Grossman School of Medicine. She's co-chair of the
14 Working Group on Compassionate Use and Pre-Approval
15 Access, an academic group that studies ethical issues
16 concerning access to investigational medical products.
17 She has published and spoken extensively on how to best
18 handle requests for non-trial access to investigational
19 drugs and related ethical issues. And she's also
20 written and spoken frequently on the history and ethics
21 of using humans in research subjects and on clinical
www.transcriptionetc.com
308
1 trial accessibility. So welcome.
2
3 ETHICAL ISSUES IN PRODUCT DEVELOPMENT AND
4 SUSTAINABILITY FOR INDIVIDUALIZED THERAPIES-DR. ALISON
5 BATEMAN-HOUSE
6
7 DR. BATEMAN-HOUSE: Hi, everyone. I want to
8 thank the FDA for the invitation to speak today and I
9 want to thank Ms. Wood for that very illuminating
10 presentation. So there's no way I can follow that, so
11 I won't even try. The best I can probably do is
12 entertain y'all with my lack of ability to manipulate
13 mechanics. So we'll see how this works.
14 So we've heard numerous people say today, you
15 know, maybe the time has come that we need to come up
16 with a paradigm shift in this drug development
17 traditional model that we've heard of. You know, this
18 pre-clinical, Phase 1, Phase 2, Phase 3, it's too slow.
19 Maybe there's situations in which we need to modify it
20 somehow. And I think that's probably true, and we're
21 here to talk about that. But before we talk about
www.transcriptionetc.com
309
1 that, I want to just remind us how we got to where that
2 is. So let's see if I can move forward.
3 Okay. So in the 1940s, we basically said, you
4 know, let's do medicine scientifically, and we're going
5 to sort of disaggregate research and treatment. And
6 they may look very similar, the same products may be
7 involved, but they have different intentions. And as a
8 result -- sorry, I'm getting distracted by -- see, I
9 told you I'm bad at technology. I should just look up
10 here.
11 They have different intentions. So they may
12 involve, you know, similar procedures. They may
13 involve similar, you know, products, but the intention
14 is different. And here's the classic definition from
15 the 80s. Research talks about a class of activities
16 designed to develop or contribute to generalizable
17 knowledge. Whereas, practice is referring to a class
18 of activities designed solely to enhance the well-being
19 of an individual patient or client.
20 So when people go into research, we often hope
21 that there will be benefit to them, but it's not the
www.transcriptionetc.com
310
1 motivating factor behind research. And this has been
2 sort of a core disaggregation that happened, like I
3 said, in the '40s and has been systematized over time.
4 With this Phase 1, this Phase 2, Phase 3 paradigm that
5 we have now really came into place in the '60s, and it
6 said, research, it is a formal thing. It has a method,
7 and it has a procedure. And it's different from
8 treatment.
9 And as a result of that we sort of came up
10 with this traditional model of who is responsible for
11 what. Patients, your responsibility is to be treated,
12 be a good patient, do what your doctor tells you to do.
13 Doctors, your responsibility is to treat patients.
14 That is your patient in front of you. It is your job
15 to advocate for them and do whatever you can do to help
16 them. Researchers, your responsibility is to conduct
17 research and to get that generalizable data that will
18 move science and knowledge forward. Research subjects,
19 you are passive. Your role and responsibility is to be
20 researched upon. Funders, your role is to figure out
21 what research is promising, to vet it, to fund it. And
www.transcriptionetc.com
311
1 then, as we mentioned earlier, typically the
2 pharmaceutical companies are the ones who have come in
3 and said, you know, “We'll take it from here. We'll
4 take the most promising research and turn it into a
5 marketable medical product.”
6 Now, this started collapsing around the '80s.
7 We had activist patients, particularly in the context
8 of HIV/AIDS who said, “I don't want to be a passive
9 research subject. I want to have agency. This is not
10 working for me anymore.” And there are many, many
11 people today who say this is not working for me
12 anymore. And we've heard examples all throughout today
13 of patients or parents who have taken science into
14 their own hands and said, “In the effort to treat my
15 child, I need to get involved in the research world.”
16 So the traditional model is evolving. Just to
17 give some examples, you know, as we've heard today,
18 research may be intended primarily as therapy. If you
19 have a research endeavor that is anticipated to help
20 one patient, that's not research for the sake of
21 research. That's research as therapy.
www.transcriptionetc.com
312
1 As a result, your research subjects, they may
2 not me acting as passive research subjects. They're
3 saying “I'm a patient. I'm here to be cured or helped.
4 This is my therapeutic option.” Patients and
5 advocates, as we just heard a stunning demonstration
6 of, they may be the ones now who are picking what
7 science to fund, what science to push forward, what
8 science to really try to get out of the realm of theory
9 and actually into the lab.
10 This means that companies may be sidelined,
11 and not necessarily because they're being pushed out by
12 patients. I don't mean that. But as we mentioned
13 earlier, if there's no market incentive, companies are
14 basically leaving the space, and that's why these
15 parents or advocacy groups are coming in. And they
16 may not be sidelined, but they're playing a less
17 prominent role. And in some cases, they're almost even
18 like subcontractors.
19 And then there's the question that, you know,
20 we bring to the FDA of, well, what and how are we going
21 to approve something out of this? I mean, typically
www.transcriptionetc.com
313
1 speaking, if it's a N of 1, like really an N of 1
2 bespoke that no one other than person with the de novo
3 mutation is gonna use, you don't need anything
4 approved. But if there's something that we can
5 extrapolate form this like a platform technology, then
6 maybe there is something we can approve. What is it
7 and how are we gonna approve it? And will something be
8 brought to the market?
9 So these are all new questions that everyone
10 in this room is currently, you know, grappling with and
11 I just wanted to lay them out. And of course, anytime
12 you shift between paradigms, it's difficult and there
13 are complications. And so some of the questions that
14 arise as we're making this shift right now is I showed
15 you that traditional outlay of roles. And one of the
16 questions is who should be playing what roles?
17 So for example, if you have a laboratory
18 scientist, are those the people that we want making
19 treatment decisions for individual patients? Maybe
20 yes, maybe no, but it's not necessarily something that
21 they've done before or been trained for. And this is a
www.transcriptionetc.com
314
1 new reality that we need to be grappling with.
2 Funding. Funding has always impacted
3 decisions as to what's going to be developed, but is
4 funding now going to be impacting treatment decisions?
5 And I'll just give you an example from a patient
6 advocacy group that I work with. They funded the
7 clinical trial, and so the expectation was, you know,
8 well, we get to pick what patients go first. Who's the
9 first to get dosed? Who's the second to get dosed?
10 Who's the third to get dosed? And are we okay saying,
11 you know, this wouldn't be happening without you so
12 okay, you get to make that decision? Or is that sort
13 of a no go request?
14 So research and therapy traditionally have
15 been distinct, and there has always been rampant
16 confusion between where is this -- I think Dr. Witten
17 says, you know, sometimes there's a fuzzy line. And
18 there's a concept called therapeutic misconception.
19 Which, I used to work in cancer, and you saw
20 therapeutic misconception all the time with patients in
21 Phase 1 clinical trials. It used to be, before our
www.transcriptionetc.com
315
1 modern era of molecularly targeted cancer, Phase 1
2 clinical trials were not intended to be therapeutic.
3 They were intended to get dosage information and to
4 move the development of a molecule along. But the
5 patients enrolled in that Phase 1 trial were really not
6 anticipated to benefit.
7 Yet, we saw time and time again that if you
8 surveyed those patients and ask why they were
9 participating in the clinical trial they would say,
10 “You know, I'm hoping to get something out of it.” So
11 there's been this ongoing confusion about, am I
12 participating in research? And if so is that therapy?
13 Is it not therapy? And it's been even more complicated
14 when the same health care provider has been the
15 investigator and the clinician. And the patient or
16 parent is like, “Well, am I being recommended to go
17 into this trial because they think it'll help me or
18 because they're the PI of this trial and they need
19 people?”
20 So this has been an ongoing issue for decades.
21 And of course, now we're getting to the situation where
www.transcriptionetc.com
316
1 research and treatment are getting even closer together
2 and, in some cases, becoming completely inseparable.
3 And that's what we're looking at in some cases with
4 some of the individualized therapeutics.
5 Although I was struck today as we heard
6 different stories from cancer vaccines to gene therapy
7 to gene editing, that there are distinctions between
8 them. You know, you can't make a one size fits all
9 statement here. But regardless, in general, we're
10 seeing the situation in these individualized
11 therapeutics where there is this sort of merger of, you
12 know, from the expectation of the participant, is this
13 research or is this treatment?
14 And that has implications for those of us who
15 are in the field. Those of -- implications for, you
16 know, the companies, the clinicians, the researchers,
17 the academic medical centers. So if something is
18 experimental but it's intended as treatment, how do we
19 handle that? Because traditionally we've had a model
20 where those have been disaggregated. And it has legal
21 implications. It has regulatory implications. It has
www.transcriptionetc.com
317
1 ethical implications.
2 And just to single out one, IRBs. So if
3 you’re doing research, you have to go through an IRB.
4 You have to get ethical review of your research.
5 That's very different from what happens in treatment.
6 In treatment, there's still the idea of you
7 need to have informed consent, but you don't go through
8 an IRB. Your paperwork that you fill out to enroll in
9 a clinical trial is very different, normally speaking,
10 from the paperwork you fill out to have, like, your
11 gall bladder removed. So you -- we have these concepts
12 that, yes, there must be informed consent. But how we
13 actually formalize them, change has been different
14 depending on which one of these realms you're in. And
15 so we need to figure out really where we are.
16 So a number of people have pointed out today
17 that some of these individualized therapies have been
18 given under the rubric of expanded access, also known
19 as compassionate use, in some cases called the eIND.
20 And I spent a lot of time in this field, so I just
21 wanted to make some comments about it. So we mentioned
www.transcriptionetc.com
318
1 IRBs a second ago. IRBs are involved in both research
2 and EA.
3 But the level of oversight is very different.
4 So if you have an IRB involved in our research, there's
5 gonna be multiple rounds of review probably, looking at
6 every line of the protocol and trying to decide how is
7 this gonna be advertised, and who's gonna be recruited,
8 and how do we make sure they understand what they're
9 getting into, and what sort of data are we gonna return
10 to them? The IRBs on expanded access basically say,
11 does this seem reasonable and is there some piece of
12 paper that we can hand somebody, either the patient or
13 the family member, to have them sign off that they
14 understand that this is experimental?
15 The levels of oversight are completely
16 different. So there are just practical ramifications
17 to trying to figure out -- we need to figure out which
18 one of these paradigms we're working in. And, just as
19 I'm saying right now, what does it matter?
20 So you know, pick a paradigm and just move
21 forward. Say it's research, say it's expanded access,
www.transcriptionetc.com
319
1 say it's treatment, say whatever, it doesn't matter.
2 It does matter because we need clarity about the
3 procedures. We need clarity both within a particular
4 institution or within a particular multi-center
5 initiative or what have you. We need to make sure that
6 patients understand what's being proposed, why and what
7 the possible risk and benefits are.
8 And of course, the way we currently do that is
9 through an informed consent discussion that is
10 memorialized with informed consent form. But we need
11 to know what that form looks like, and we need to try
12 to choose which one of these to use to do that
13 properly. And we need to make sure that the
14 stakeholders involved understand their
15 responsibilities.
16 So I just wanted to go briefly over this. And
17 expanded access when you're talking about, you know,
18 sort of, like, your classic idea of an unapproved drug,
19 the request to do, you know, use this unapproved drug
20 outside of a clinical trial has to be initiated by the
21 physician. Of course, the patient may be the one who
www.transcriptionetc.com
320
1 initiates this conversation with the physician, but you
2 have to have the physician on board. And the physician
3 is the one who is in charge of reaching out to the
4 company who is developing this new product and saying,
5 “May we please use this outside of the clinical trial.
6 Here is the patient, here's why I want to use it, and
7 here's why I can't use it in a trial.”
8 And if the company says no, that's basically
9 the end of the story. This is where we see social media
10 campaigns and whatnot trying to make companies change
11 their minds. But generally speaking, if a company says
12 no, that's the end of the story.
13 If a company says yes, this is where the FDA
14 gets involved to look over the proposal, make sure that
15 there's no obvious safety concerns, see if there are
16 any amendments that they think need to be added to the
17 proposal. And of course, this is where the IRB gets
18 involved. So this is your sort of ladder that you have
19 to go through to do expanded access for a single
20 patient.
21 And I just want to note that really the
www.transcriptionetc.com
321
1 gatekeeper in this situation is the company. Again, if
2 the company says no, that's basically the end of the
3 story. The FDA cannot say to the company, “You must do
4 this.” The IRB cannot say to the company, “You must do
5 this.” It's the company that's the gatekeeper. And I
6 want to point this out because, in the situations that
7 we've been hearing today, the roles of companies have
8 been changed, if not minimized or completely removed.
9 And we're really talking about things that are
10 happening in the academic center.
11 So is the role of gatekeeper now being taken
12 over by this investigator, and, if so, does that
13 investigator know that they are now the gatekeeper? Do
14 they want to be the gatekeeper? Are they comfortable
15 being the gatekeeper? You could say there is no
16 gatekeeper; everyone gets what they want. But I don't
17 think that's a sustainable model, and we probably
18 shouldn't advocate it.
19 But if the investigator doesn't want to be the
20 gatekeep do we say, “Okay, FDA, now you're the
21 gatekeeper? Every time someone wants to use one of
www.transcriptionetc.com
322
1 these individualized products outside of a clinical
2 trial, you really need to do the due diligence and
3 decide yes or no.” And it's up to the FDA to decide if
4 they're comfortable with that role or not, but that's
5 an additional burden over what they've currently been
6 asked to assume.
7 Or we could say to the IRB, “Hey, before
8 you've kind of done a rubber stamp. You know you've
9 looked at this and said, ‘Is it reasonable?’ and okay
10 make sure that there's a piece of paper for someone to
11 sign. But now we're in a new paradigm, and you really
12 need to be involved.” I don't know. Any of these are
13 possible, but there's a question. And we need to
14 figure out what to do.
15 I just want to point out that I do have a
16 concern about the idea of this investigator being the
17 gatekeeper, although it might seem like the obvious
18 choice, simply in that I'm concerned that having
19 proximity to a patient might make that investigator
20 make decisions that are a little bit too close for
21 comfort. And the sort of paradigmatic example I'm
www.transcriptionetc.com
323
1 thinking of is if you have a patient who is
2 deteriorating in front of you. Is there some point
3 where the investigator would say, “Gosh, in an ideal
4 world we would, you know, do some more work on this?
5 I'm not 100 percent comfortable, but we can't wait
6 anymore.” And maybe that's okay if it really is an N
7 of 1 and this is the only patient that it's gonna
8 impact.
9 But I don't think that's okay if there are
10 other patients out there that we're gonna be trying
11 this intervention on, and hopefully we'll be collecting
12 data from this first experiment. And to the idea that
13 someone might be -- jump the gun a little bit, that
14 makes me concerned. And of course, there's also the
15 point where if that patient or that patient's family or
16 that patient's community is the one funding the
17 investigator, is there even more of this potential
18 conflict of interest?
19 And, you know, if you can think of these
20 problems happening theoretically then they're probably
21 gonna happen in real life. And so is there something
www.transcriptionetc.com
324
1 we can do to prevent foreseeable issues? Say, we're
2 fine with the investigator being the one making these
3 decisions, but let's come up with some rules of
4 engagement.
5 So just to repeat, we need to decide is this
6 research? And if it is research, we need to modify the
7 way that we think about these. And we need to train
8 IRBs so they understand what they're looking at. And
9 we need to come up with some way of deciding what
10 experimental procedures we really are happy with and
11 whatnot. Or if we say this is clinical care, fine.
12 That's fine. I'm okay with that, but we need to,
13 again, come up with rules of engagement.
14 And one of the questions I have is, you know,
15 do we say any licensed M.D. in the country should be
16 able to do this or only certain M.D.s, only in certain
17 settings, only with certain oversight? And if we are
18 gonna say there are some sort of rules of engagement,
19 who's gonna develop them and who's gonna enforce them?
20 All right. So then to get into the bread and butter of
21 ethics concerns I always have to talk about justice.
www.transcriptionetc.com
325
1 So one of the things that I'm concerned about
2 here -- and this is not to say we should not do this.
3 I absolutely think we should do this, but these are
4 questions we need to ask. How do we justify extensive
5 use of resources to benefit only one person or maybe
6 very few people?
7 And I think everyone in the room today has
8 been hitting upon the same theme of, you know, we
9 collect data from those N of 1s, and we find some way
10 to have that data push us forward in ways that will
11 help more patients. So I think that's great that we're
12 all on the idea of leveraging findings to help wider
13 numbers of patients. But we need to figure out how to
14 do that. We must plan for it, and it must be something
15 that we say is non-negotiable, not something that, you
16 know, after we do some cases, we'll figure out how to
17 go back and do a reanalysis of the data. I don't think
18 that's acceptable. I think we need to say right from
19 the start, as we're building our plans, here's how
20 we're gonna do this.
21 Enhancing access, decreasing obstacles, and
www.transcriptionetc.com
326
1 costs. I think just last week there was a meeting,
2 maybe here at FDA -- I don't know -- for Rare Disease
3 Week. And there was a panel on individualized
4 therapeutics, and there was a patient who had received
5 phage therapy. And she talked very movingly about how
6 she was receiving care in Richmond, Virginia. And she
7 had a very bad intractable infection and came up with
8 the idea of phage therapy.
9 And her physician either was unable or
10 unwilling to do it, and so she had to go to Yale to get
11 access and how she literally thought she might die on
12 the train because it was just too much to ask for her
13 to do it. And she did it. And it worked, and that's a
14 success story. But it's also a cautionary tale because
15 I'm deeply concerned about the idea that we are, you
16 know, not thinking about access from the get-go and
17 trying to figure out how to minimize those barriers as
18 much as possible.
19 We live in a country that has unjust access to
20 health care, and so I understand when I'm saying we
21 need to enshrine justice as a core principle in access
www.transcriptionetc.com
327
1 to research that seems a little bit odd because we
2 don't have it in access to just normal clinical care.
3 But I'm aspirational like that. What can I say?
4 You know, so as we're building a phage bank or
5 as we're developing consortium, I just am encouraging
6 people to invest the time, the money, and the planning
7 up front to try to figure out how to minimize barriers
8 to patients both in real time and downstream. And
9 also, I have an international point here -- I'm not
10 sure we've talked internationally yet today -- about is
11 it possible to harmonize regulations now from the get-
12 go so that once something is successful in the United
13 States we can very easily translate it to, say, Canada
14 or other countries where people will be wanting to try
15 these?
16 More justice concerns. We've talked about
17 this numerous times today in terms of how do we decide
18 how much information we need to acquire before we use a
19 product on a patient. And, you know, I don't have a
20 magic bullet answer for this. This is an ongoing
21 question anytime you have a novel intervention. We're
www.transcriptionetc.com
328
1 trying to balance earlier access and the possible
2 benefit it offers with the risk that comes with getting
3 something, you know, earlier before all the testing has
4 been done.
5 So my only recommendation here is that we need
6 to embed frequent evaluations of safety. It can't be
7 something where we do, you know, we're gonna do 20
8 people, and then we're gonna do a, you know, a post hoc
9 analysis. I think we should be looking at this in real
10 time as we go along and manage risk cautiously.
11 And then, last but not least, I just wanted to
12 say, the thing that really gives me heartburn about all
13 of this is that, as we are doing this novel, like,
14 amazing science for certain patients, there are other
15 patients who are being told, “I'm sorry you can't get
16 access because it hasn't gone through a mouse model
17 yet, or it hasn't gone through a primate model yet.” Or
18 “the Phase 1 was only enrolling 20 patients, and
19 they've already enrolled those 20 patients. so you're
20 gonna have to wait until a Phase 2 trial opens.” It is
21 hard for me to understand how we are gonna justify to
www.transcriptionetc.com
329
1 other patients who are not in this, like, exclusive,
2 individualized therapy category about why they still
3 need to observe a status quo and rules that are, you
4 know, being turned and modified in this particular
5 context.
6 So that's the thing that keeps me up at night.
7 And in the meantime, the thing that makes me really
8 happy is the fact that we're having, like, amazing
9 science that's gonna help patients. So I want the
10 amazing science to help patients to go forward, but I
11 want us to be aware of all the challenges that it
12 presents for us and to be proactive about addressing
13 them. Thank you.
14 DR. WITTEN: I'd like to next introduce Dr.
15 P.J. Brooks. He's a program director at the NCATS
16 Office of Rare Diseases and Research. He received his
17 Ph.D. in neurobiology from the University of North
18 Carolina at Chapel Hill and completed a postdoctoral
19 fellowship at the Rockefeller University. Since
20 joining NCATS and the Office of Rare Diseases, Dr.
21 Brooks has been working on accelerating clinical trials
www.transcriptionetc.com
330
1 in rare diseases by moving beyond one disease at a time
2 approaches. Examples include the development of
3 therapeutics that target shared molecular mechanisms,
4 underlying multiple rare diseases, platform
5 technologies for delivery of nucleic acid therapeutics,
6 and the implementation and recommendations from the
7 NCATS Cures Acceleration Network regarding the
8 acceleration of gene therapy clinical trials. Thank
9 you.
10 BEYOND ‘ONE DISEASE AT A TIME:’ ACCELERATING CLINICAL
11 TRIALS OF GENETIC THERAPIES BY GROUPING RARE DISEASE
12 PATIENTS ACCORDING TO UNDERLYING DISEASE MECHANISM -
13 DR. PHILIP J. BROOKS
14
15 DR. BROOKS: Thank you Celia and thank you to
16 my FDA colleagues for the invitation to participate in
17 this meeting. It's really exciting and it's something
18 that we think about a lot. And I'm very happy to
19 participate. And I'll be focusing on, more generally,
20 the idea of trying to go beyond one disease at a time
21 and how we can accelerate clinical trials of genetic
www.transcriptionetc.com
331
1 therapies by grouping rare disease patients according
2 to underlying mechanism.
3 This is my standard federal government
4 disclosure slide. And here you see the basic problem
5 that I think we're all trying to address is the rapidly
6 increasing number of disorders with a known molecular
7 basis, due in large part to DNA sequencing. And this
8 is likely to continue. And the big problem is that, at
9 the present time, we've only about 600, 500 or 600 or
10 so with therapy. And at the rate we're going, as my
11 director, Chris Austin, said, it's gonna take about
12 2,000 years to get treatments for every one of these
13 diseases, and that's just too darn long. And if we're
14 gonna do something about this we don't need sort of
15 minor tweaks to the process. We need some pretty
16 fundamental changes in the way we think about these
17 diseases and the way we design clinical trials.
18 And there's another aspect to this as well
19 that sometimes doesn't get appreciated, but I think
20 Jill Wood kind of hit on it. And that has to do with
21 this issue of the different types of rare diseases. So
www.transcriptionetc.com
332
1 this is a slide taken from a recent publication by
2 Orphanet, and we're looking at, within rare disease,
3 there's different prevalence, you know, the highest and
4 the lowest. And if you're thinking about the
5 percentage of all rare disease patients, the majority
6 of them are these high prevalence diseases, right?
7 That makes sense, about 70 percent of the patients.
8 But if you think about the number of diseases, it's a
9 very small fraction. Way on the other side, the low
10 prevalence diseases, the Sanfilippo, you know, III C,
11 D, et cetera, you've got a very small number of
12 patients but about 3,000 or so diseases.
13 And then, if we're gonna develop these
14 treatments according to a standard model, the diseases
15 of commercial interest are these ones for obvious
16 reasons. But who's ever gonna do anything about these
17 diseases here, these 3,000 diseases? If we're gonna do
18 this one at a time I'm not sure we're gonna get to any
19 of them. So we have to really reevaluate how we
20 approach this problem, particularly because, as I'll
21 get to later, when we're talking about some of these
www.transcriptionetc.com
333
1 diseases and monogenic diseases, we really do have
2 therapies and treatments that have a pretty high prior
3 probability of success. Oops. Wrong way. Okay.
4 So I think it comes down to the old lumpers
5 and splitter distinction that we're all familiar with
6 from different facets of life. This was taken from a
7 paper written by Victor McKusick, the famous
8 geneticist, many years ago. And he was talking about
9 limpers and -- lumpers and splitters in the context of
10 nosology. As you can tell by the way he drew these
11 different pictures, he was a big fan of the splitters
12 and not so much the lumpers. And maybe that made sense
13 for the point he was trying to make; I honestly don't
14 know. But with all due respect, I think in this day
15 and age we've got to be lumpers wherever we can. And
16 we've really got to focus on the commonalities across
17 diseases rather than what makes them different.
18 And the really, I think, wonderful opportunity
19 here is in the area of monogenic diseases. And you
20 could say this is perhaps the biggest lump of all, at
21 least within personalized or individual therapies.
www.transcriptionetc.com
334
1 Because, when you talk about monogenic diseases, these
2 are diseases that relate from mutations in a single
3 gene. And that means we know what the problem is, and
4 we know at least what some solutions are, which is
5 quite different than many other diseases.
6 So we have gene therapy which could deliver a
7 normal version of that mutant gene into the relevant
8 cell types of the patient and then, more increasingly
9 now, genome editing where you can -- need to deliver
10 genome editors or enzymes into these cell populations
11 to correct that disease-causing mutation in the
12 patient's cells. Particularly here we're talking about
13 somatic cells. That's what we focus on in the United
14 States. We're not -- we don't do germline editing in
15 the United States.
16 And so the idea then -- so the real challenge
17 for both of these is to deliver these treatments to
18 enough target cells at the right time in development of
19 disease progression to potentially treat, cure, or even
20 prevent some of these diseases. And this is really
21 fundamentally true for all monogenic diseases. And the
www.transcriptionetc.com
335
1 good news here is that we've actually made some
2 progress in being able to do this. As Guangping Gao
3 talked about adeno-associated virus or AAV, these
4 really are effective vectors to deliver genes into
5 cells.
6 They have an excellent safety record in humans
7 to date, clinical success stories to approved products,
8 and we see a lot of pre-clinical success stories. If
9 you go to the American Society for Gene and Cell
10 Therapy meetings, there's a lot of people curing a lot
11 of mice of a lot of genetic diseases. It works quite
12 well. But when you go to develop these into the
13 clinic, we run into this one disease at a time
14 approach.
15 And quite often you see the focus on the more
16 common rare diseases. And this is slow, inefficient
17 and results in duplications of efforts across different
18 programs. It costs animals, time, and money and
19 particularly the time and money in some cases of the
20 parents who are trying to develop these therapies, of
21 which there isn't very much. And there's this obvious
www.transcriptionetc.com
336
1 bias towards the more common rare diseases. So what
2 makes sense would be to do -- to start these clinical
3 trials for multiple diseases at a time using the same
4 platform vector and that should increase the efficiency
5 and reduce the time of clinical trial start up, would
6 make sense.
7 So we're gonna try to test that specifically
8 in a program we call the Platform Vector Gene Therapy
9 or PaVe-GT project. And this is a collaborative effort
10 between our office, Office of Rare Diseases Research at
11 NCATS, and other collaborators at NCATS, in particular
12 the Therapeutics Development Branch led by Don Lo and
13 strategic alliances led by Lili Portilla, as well as
14 colleagues from NHGRI and NINDS who will be working
15 with us on the clinical trial. Do you have some water?
16 So to be clear, this is a pilot project where
17 we're gonna be doing essentially a public platform
18 vector gene therapy trial at the NIH Clinical Center
19 involving all investigators from NIH. And the idea is
20 to move forward with -- ultimately, towards clinical
21 trials for gene therapy for four rare genetic diseases
www.transcriptionetc.com
337
1 together, each of which are of no commercial interest.
2 So these are -- these -- the very far end of that graph
3 I showed before. We use the same AAV vector, the same
4 route of administration, the same serotype, use the
5 same production purification methods because we've
6 heard from our FDA colleagues many, many times that the
7 process is the product. And the only thing that will
8 be different are the therapeutic gene constructs for
9 the different diseases. And the question is to what
10 extent can we increase the speed and efficacy of
11 clinical trials startup by really trying to maximize
12 this explicit platform vector-based approach.
13 But the other thing I think that's gonna be
14 different about this is that we intend to do this
15 publicly and make all of the data, including the
16 biodistribution data, the toxicology data, all of our
17 communications with the FDA up to and including the IND
18 submissions that will hopefully get approved -- we
19 intend to make that public and publicly available so
20 that all these documents and things can be used by
21 others perhaps even in a cut and paste manner. And in
www.transcriptionetc.com
338
1 doing this and thinking about this approach, honestly,
2 we had in mind people like Jill and all the parents
3 that we meet who are trying to figure out how to do
4 this by themselves. You know, it's got to be hard
5 enough having a child with a rare genetic disease.
6 But then to expect them to become
7 entrepreneurs and drug developers, and we think how can
8 we provide help for some of these individuals? And I
9 wish there was more we can do, but this is one approach
10 that we think has potential benefit. And essentially,
11 the idea here is that if we're going to do these one
12 disease at a time for each vector we would make, for
13 each gene, for each disease, we go through each of
14 these steps in parallel, one right after the other.
15 And that takes time and money, and time is an issue.
16 But the question is, if you group them all together,
17 can we utilize the fact that we're doing everything in
18 the same pathway to avoid having to do, perhaps the
19 biodistribution studies or some of the other steps, and
20 reduce the amount of time to clinical trial startup?
21 And so here's sort of where we are on this.
www.transcriptionetc.com
339
1 We've got the collaborating investigator Carsten
2 Bonnemann from NINDS, and Chuck Venditti from NHGRI.
3 And we'll be working with two neuromuscular diseases
4 and two rare organic acidemias using AAV9 for all four.
5 And we're undergoing -- proof of concept studies and
6 mouse models in human cells.
7 And I say we're gonna let you know about our
8 communication with the FDA, and I can tell you that we
9 had a communication with the FDA a few months ago and
10 talked to them about this. And it went quite well,
11 better than I kind of anticipated. They were quite
12 good about the idea and I think were all supportive
13 about being as transparent as possible. And we’ll be
14 anticipating our initial INTERACT meeting with the FDA
15 later this year. And the key challenge we're facing,
16 which I guess is one that everybody is facing and
17 probably led to the meeting we had a few weeks ago, is
18 how do we get AAV vector made for the clinical trials?
19 Because we run into the same problem that everybody
20 else does. So we are working on that as well.
21 And then earlier, Peter Marks, mentioned the
www.transcriptionetc.com
340
1 effort that we've been involved with working with them,
2 as well as the FNIH, to develop a more broad public-
3 private partnership for some of these individualized
4 therapies and specifically focusing on AAV gene
5 therapy. And it's a pleasure to be working with them
6 and look forward to continue doing so. And I just
7 wanted to point out that there are some actual
8 parallels between our PaVe-GT effort and this other
9 public-private partnership that is in progress that we
10 hope can ultimately be leveraged.
11 In both cases, there has to be some decision
12 about the serotype that's going to be used for the
13 different diseases. We're choosing a single one.
14 Perhaps there'll be multiples here. I put question
15 marks on all these because there's still some questions
16 about how our -- this public-private partnership is
17 gonna work, but I think there's some clearly -- clear
18 issues that'll need to be addressed. We're doing all
19 of our work at the NIH Clinical Center.
20 I might point out, in part because we want to
21 utilize that resource, but also because we're so
www.transcriptionetc.com
341
1 committed to making all of the data publicly available
2 that we felt that the most efficient way to do it is by
3 having everyone be a government employee. If we had a
4 commercial company involved or even academic medical
5 centers, we might have to deal with some of the
6 intellectual property issues. So having it all done
7 within the NIH kind of avoids that problem.
8 In -- oops. But ultimately in this effort we
9 might be involving many clinical sites. As I said,
10 we're gonna make all of our communications public. And
11 what we had kind of hoped, actually, is that some of
12 these documents and things might ultimately spill over
13 and benefit this potential public-private partnership.
14 The single manufacturer is, of course, key.
15 But one might consider a consortium here, and,
16 you know, we chose four rare diseases for specific
17 reasons and largely due to the availability of diseases
18 and the investigators at the NIH. But here in this,
19 whatever, effort, I think a big question is gonna be
20 how do we choose and determine what diseases that would
21 be under consideration? But I think both of these
www.transcriptionetc.com
342
1 things are kind of moving forward towards a day in
2 which we can really make access to these treatments a
3 lot more available for a lot -- much larger numbers of
4 patients and families and hopefully take the burden off
5 of people like Jill and other parents.
6 So then also the next phase, if you will, of
7 the way to treat genetic diseases is genome editing.
8 And I just want to briefly touch upon a program that
9 we're involved within at NIH. This is a program funded
10 by the NIH Common Fund, part of the Office of the
11 Director. And it's on somatic cell genome editing.
12 And it's coordinated by NCATS, Chris Austin, myself,
13 and in association with many other program directors
14 across the NIH.
15 And the goals of this program are to lower the
16 barriers for new genome editing therapies by testing
17 genome editing reagents and delivery systems and better
18 animal models. These are not specific disease models
19 but rather animal models created to allow us to detect
20 genome editing in different cell types to maximize a
21 broad utility, a big focus on testing unintended
www.transcriptionetc.com
343
1 biological effects. And I should say, unintended
2 biological effects specifically in human cells and
3 human cell systems for the reasons that Peter
4 mentioned, that the human genome is special. We also
5 have some interest in monitoring these cells in vivo.
6 We have some -- the biggest focus of the
7 program is finding ways to deliver genome editors to
8 different cell types. There's also a small part on
9 increasing the genome editing repertoire, sending
10 genome editing enzymes, and of course a coordinating
11 center. And to give you a sense of the breadth of the
12 program, this is the number of awards. The total
13 budget of the program is around $180 million. And you
14 can see that by far the majority of the awards --
15 almost half of them are focused on the delivery systems
16 because that, as we see, is the biggest challenge.
17 There are some cells and tissues we can deliver to
18 pretty well but many that we can't at all, and that's
19 really what the focus is on.
20 And if you want to learn more about it, here's
21 the website. And the way we see this focused --
www.transcriptionetc.com
344
1 getting into the IND enabling process is kind of
2 illustrated here, that this program will not
3 specifically be funding any clinical studies, but
4 rather what we think about is filling gaps. And there
5 might -- one gap might be for a specific disease a need
6 to be able to deliver genome editors to a particular
7 cell type. But another gap actually that we focus on,
8 and would really like to have some impact on, is the
9 gaps in the regulatory process. That would allow our
10 FDA colleagues to be able to regulate these products
11 more effectively. And indeed, when we were developing
12 this program and as we are going through it, we have
13 close communication with people in FDA CBER to try to
14 maximize that potential.
15 So finally, let me turn to addressing the
16 question that FDA asked us to address which is what
17 kind of opportunities and possibilities might work in
18 the future to make some of these approaches scalable
19 and sustainable. And I have some thoughts about that.
20 So one would be to adapt the approach they use now in
21 the development of vaccines, when one is producing a
www.transcriptionetc.com
345
1 new vaccine using a new strain. I should say that this
2 is not an idea I came up by myself. This was
3 originally brought to me by one of my colleagues, Mike
4 Cirillo at NCATS, and I've heard Peter talk about it
5 also.
6 But the basic idea is that, when someone's
7 first gonna set up a vaccine production facility, the
8 FDA would review -- would have to consider all of the
9 aspects of setting up the process: the manufacturing
10 facility, CMC potency assays, all kinds of things in
11 addition to specific strains. But once a system is
12 ongoing and producing, if from one year to the next
13 you're just simply switching strains, then the FDA
14 review can just focus on what's different, what's new,
15 which in this case would be the new strain. And I'm a
16 little imprecise about this because I've never
17 regulated a vaccine, but I think you kind of get the
18 idea.
19 And so we could take the same basic principle
20 and apply it to gene therapy. When you're originally
21 developing a gene therapy product, the FDA's gonna have
www.transcriptionetc.com
346
1 to consider all these things, the manufacturing and the
2 whole, sort of, regulatory process. But the idea and
3 the hope is that once we're making the same vector, and
4 in the same manufacturing facility and holding all
5 these things the same and simply changing the
6 therapeutic transgene, the FDA review can focus on
7 this. And that could streamline the process. And not
8 surprisingly, that's basically the idea we're trying to
9 test in PaVe-GT.
10 But I think the real exciting option has to do
11 with genome editing. And late last year this
12 publication came out from David Liu, who is funded by
13 our consortium in part, a new genome editor that can
14 carry out editing without creating double strand
15 breaks. And the notable thing about this prime editing
16 effort -- this prime editing enzyme is that this single
17 enzyme could, in principle, correct almost 90 percent
18 of known genetic disease-causing mutations. And that
19 really seems like a potentially exciting platform.
20 And so if you imagine -- and again, obviously,
21 I'm thinking about the future here. These are perhaps
www.transcriptionetc.com
347
1 forward-thinking statements. But if you think about
2 the production utilization of such a biologic, you can
3 end up with a single biologic that would be of
4 potential therapeutic relevance to almost 90 percent of
5 genetic diseases.
6 And once that would be approved, then, if you
7 just want to add additional diseases to it, the only
8 things that would be different are these guide RNAs
9 that would direct the location of the editor within the
10 genome. Guide RNAs of course are oligonucleotides, and
11 oligonucleotides are regulated by the Center for Drugs.
12 But the good news is, of course, a lot of excitement
13 going on in this area with the oligonucleotide
14 therapies for the rare genetic diseases.
15 And I'm sure you've all heard about the work
16 by Tim Yu on the development of Milasen, and obviously
17 the FDA, CBER, and CEDR are able to communicate on
18 this. So it doesn't seem like an insurmountable
19 problem. And so if you get back to this sort of
20 optimized situation in the future, we’d be looking at
21 perhaps a single biologic. And then the guide RNAs
www.transcriptionetc.com
348
1 would be assessed under really a streamlined process
2 that takes into account the platform capacity of
3 oligonucleotides to be able to see them as a class of
4 molecules that would optimize the toxicology assessment
5 of those as well. And, you know, thinking about the
6 future of treating monogenic disease, this seems to me
7 to be at least an aspirational idea of where we might
8 want to go.
9 And so I think I'll just kind of stop there
10 and summarize that, for monogenic diseases, gene
11 therapy and gene editing have clear and obvious
12 therapeutic potential for many monogenetic diseases.
13 This one disease at a time approach that we're doing
14 now is not going to address these low prevalence
15 diseases of no commercial interest, despite the fact
16 that the biological rationale of treating those
17 diseases is just as good as the common diseases. And
18 that does not seem acceptable. That's a major reason
19 why we need to do something different. And so we need
20 radically different types of clinical trials and
21 regulatory platforms to bring gene therapy and gene
www.transcriptionetc.com
349
1 editing therapies to all the patients who might benefit
2 from them.
3 Oh, and just one last thing. I think we've
4 been always talking about individual therapies, and I
5 understand why we're saying that. But I was telling
6 you at the beginning, there's lumpers and there's
7 splitters. And when you talk about individual diseases
8 you're focused on splitting.
9 And I think -- I happened to come across this
10 from a publication in Stat News. And I like the idea
11 of industrializing personalization because I think
12 that's sort of what we're talking about, how to take
13 making individualized therapies into some sort of an
14 industrial process. So I guess I'll leave you with
15 that as well as our NCATS contact information. Thank
16 you very much.
17 PANEL DISCUSSION WITH Q&A
18
19 DR. WITTEN: Okay. Thank you. I'm gonna ask
20 the speakers and the -- to take their seats at the
21 panel. And also, I'm going to ask Dr. Chip Schooley
www.transcriptionetc.com
350
1 and Dr. Julienne Vaillancourt. Dr. Schooley was
2 introduced in the last session, but I just want to
3 introduce Captain Julie Vaillancourt, whom many of you
4 know. She's an officer in the U.S. Public Health
5 Service, and she's the Rare Disease Liaison for the
6 Center for Biologics Evaluation and Research and
7 coordinates our rare disease program.
8 Thank you for joining. We'll take questions
9 from the audience, but, in the meantime, I have a few
10 questions to start off the discussion. I’d like to ask
11 the two panel members who just joined us first. So
12 I'll start with Julie. This session has been about
13 collaborations, ethics, and stakeholder roles. And I
14 wonder if you can comment on collaborations at FDA,
15 CBER, and the rare disease area that supports
16 product development?
17 CAPT. VAILLANCOURT: Absolutely. Actually, we
18 have a really rich -- oh, thank you. At CBER, we have
19 a very rich collaborative environment when it comes to
20 our focus on advancing development of biological
21 products for rare diseases. We have collaboration
www.transcriptionetc.com
351
1 internally in CBER. We actually, as part of our rare
2 disease program, have a Rare Disease Coordinating
3 Committee that meets on a monthly basis. It's
4 comprised of representatives from each of the offices
5 in CBER: our product offices, our Office of
6 Epidemiology and Biostatistics, our -- and others.
7 And also, we collaborate extensively across
8 the agency with the Office of Orphan Products
9 Development, OOPD, and with CDER's rare diseases
10 program. And I also want to say a newer entity in the
11 last two or more years is the Patient Affairs staff in
12 the Office of the Commissioner. And they are very
13 instrumental in helping to facilitate making sure that
14 the patient voice is heard and that there are ways and
15 many mechanisms to engage patients with each of the
16 centers. And we continue to work with these different
17 groups across the agency, and we're developing new
18 collaborations every day.
19 Since I've been the rare disease liaison in
20 September of 2015, I've been participating in a Rare
21 Disease Council that is headed by the Office of Orphan
www.transcriptionetc.com
352
1 Products. And again, it's an effort that brings
2 representatives that work in -- on rare disease focused
3 development from across the agency. We meet
4 periodically. We share information, best practices.
5 Sometimes we bring in outside speakers. There's also a
6 new Rare Disease Round Table that was started by
7 Theresa Mullen in the Office of Drugs that is engaging
8 some outside stakeholders as well.
9 We have -- I'm going on and on, but you can
10 see it's very rich. We also coordinate and have a rare
11 disease cluster with our European colleagues. It's a
12 rare disease cluster that is headed by CDER's Rare
13 Diseases Program, but a number of us from CBER
14 participate. And sometimes we are asked to have
15 discussions with CBER regulated products and issues
16 with our EMA colleagues. And Health Canada has more
17 recently joined in those monthly discussions.
18 And we also coordinate with our outside
19 stakeholders, our external stakeholders, such as the
20 National Organization for Rare Disorders, NORD. We
21 have been part of their planning committee for a number
www.transcriptionetc.com
353
1 of years. And we also have some newer cooperative
2 agreements with them like our -- someone from our
3 Office of Biostatistics and Epidemiology is working on
4 a collaborative project with NORD and with a patient
5 advocacy group on a natural history study and the use
6 of a mobile app.
7 So there's lot of exciting work going on. And
8 I'll have to say we're making room for the whole topic
9 of today's meeting of individualized therapeutics. And
10 it's, for example, we've mentioned -- a few people have
11 mentioned today about the rare disease meeting that
12 took place last Monday or -- yeah, it was last Monday.
13 It's gone by so fast.
14 Anyway, that was headed by Dr. Janet Maynard
15 from the Office or Orphan Products. However, it was a
16 collaborative team effort, and there was representation
17 from CBER and from CDER. And the afternoon was all
18 focused on individualized therapeutics. And, you know,
19 there are other examples, so I hope I've given you a
20 sense of the breadth and depth of collaboration that
21 CBER is involved in, as we really work toward the
www.transcriptionetc.com
354
1 development of biologics for patients with rare
2 diseases.
3 DR. WITTEN: Okay. Thank you. I'm gonna take
4 a question from the audience.
5 MS. HESTERLEE: Hi, so it's Sharon Hesterlee
6 from the Muscular Dystrophy Association. So I'm very
7 interested in this problem of the ultra-rare diseases.
8 At MDA, you know, we hear a lot about Duchenne and SMA
9 and ALS, but the majority of the over 45 diseases we
10 cover are ultra-rare. So this is a problem that I
11 think about a lot and that keeps me up at night. And
12 I've also spent the last three-and-a-half years working
13 in industry in gene therapy and heading gene therapy
14 projects, in charge of budgets for gene therapy
15 projects. So this is kind of comments for P.J.
16 I'm really interested in this effort to create
17 a platform approach. But the caution I would give you
18 is that this idea of using the same serotype, or the
19 same capsid and vector, the same route of
20 administration, you know, same manufacturing
21 techniques, these are things that companies are already
www.transcriptionetc.com
355
1 doing. I mean, they're already doing that to try to
2 get those economies of scale. And I'm gonna tell you,
3 your savings in money are very, very small. Savings in
4 time, also pretty small. So I don't know that this
5 platform approach and this idea of this platform
6 approach at the pre-clinical, early clinical phase is
7 really gonna be that big of a time or cost savings.
8 I do think where you could see more savings
9 are doing things like doing a platform trial approach.
10 The problem is these transgenes matter, so there's only
11 so much you can do to sort of combine your efforts at
12 the pre-clinical stage. You can have tox related to
13 different transgenes and expression of transgenes. I
14 just wanted to make that point that I think companies
15 are already doing this.
16 They're already standardizing those things.
17 They're already standardizing their manufacturing
18 assays. They're doing all of those things. It's not
19 enough. Your budgets are still $20, $30 million, even
20 for an ultra-rare disease. So I think it's just
21 something to consider that that may not be as tractable
www.transcriptionetc.com
356
1 a target in trying to reduce costs.
2 DR. WITTEN: Thank you. Are there comments?
3 Do you have comments on this?
4 DR. BROOKS: Yeah. I think -- I'm sure there
5 are companies doing that, but I think that information
6 isn't made available. So we would want to make it
7 available to the whole community so it's not kept
8 within a company. And I think that certainly there
9 are many cost drivers in this, and we don't expect that
10 this effort by itself will, you know, reduce that.
11 I think our focus of the AAV manufacturing
12 meeting that we had was to try to find ways to reduce
13 the cost. But improving not just even the pre-clinical
14 stuff, but even as we think about the clinical trial
15 design within PaVe-GT, I do think we'll also be looking
16 for ways to increase the efficiency of the process.
17 But to try to learn what we can do and also to make it
18 public so everybody can benefit from it is a major
19 aspect of this goal. But certainly there's -- there
20 are -- there's more to do than that. I would agree.
21 DR. WITTEN: Thank you. I'd like to address a
www.transcriptionetc.com
357
1 question to Dr. Schooley. So Dr. Schooley, this is
2 about developing products. So you described in your
3 talk the collaboration of iPATH and of the phage
4 referral network so that, if a patient came in need of
5 a treatment, the group would collectively search their
6 inventory to see if there was phage available that
7 would benefit them. But that that only -- that was not
8 true for the -- I think you said it was the minority of
9 the patients who came that you were able to find
10 something. So I'm wondering what type of collaboration
11 or what type of effort do you think would be needed and
12 by whom in order to be able to develop phages for -- so
13 that no infectious disease is left behind, so to speak?
14 DR. SCHOOLEY: Over the short term, the
15 problem is the biology. You actually physically have
16 to have the organism in -- the bacterium in the same --
17 on the same plate or in the same liquid medium as the
18 phage candidate you want to use, which requires you to
19 have -- to disseminate that organism to whatever labs
20 or groups have libraries of phages that target that
21 organism. So right now, the limitation is how many
www.transcriptionetc.com
358
1 groups have large enough libraries that you can
2 practically get a given patient's organism out in time
3 to be able to help the patient.
4 Having more comprehensive libraries that could
5 be screened more easily would be act- -- and methods by
6 which you could screen them quickly would be great.
7 Down the longer term, if you had -- if we could by AI
8 learn to predict from AI what bacteria could be
9 attacked by which phages, you could actually do it with
10 whole genome sequencing. We're a long way from that
11 because there are more variables than equations these
12 days. But that would be the -- down the road, I think,
13 a very important approach.
14 One could also envision situations in which
15 phages were engineered to have a broader host range, in
16 essence become more like antibiotics and less like
17 phages. What you would be giving up there is the
18 specificity of phages from the standpoint of the
19 microbiome and the other advantages of the laser like
20 approach. And you would also begin to see, if you had
21 widespread use of phages with engineered phage with
www.transcriptionetc.com
359
1 broad host range, they would behave like antibiotics do
2 in the hospital. You would begin to have phage
3 resistant organisms that would then behave the same way
4 antibiotic resistant organisms do.
5 So I think, over the short term, larger phage
6 banks that could be searched more easily and production
7 facilities that were able to take that burden off the
8 hands of academic laboratories and produce phages in a
9 more standardized way would increase the throughput.
10 The sources of funding for those really haven't yet
11 been identified.
12 DR. WITTEN: Thank you. Question from the
13 audience?
14 MR. THAKUR: Yes, I'm Neil Thakur from the ALS
15 Association. So I had a question for P.J. about the
16 model that you're talking about and the ultimate vision
17 of success. And so I think what you were saying is the
18 idea is that you would get a manufacturing process
19 approved by the FDA. And when it comes time to bring
20 on a new disease or a new application of the
21 technology, the FDA review would be expedited. And so
www.transcriptionetc.com
360
1 what I'm trying to understand then is does that mean
2 that the clinical center or the facility that's doing
3 all this manufacturing would then become the hub for
4 AAV9 for these ultra-rare applications? Or would
5 somehow -- could this model be expanded or exported to
6 other facilities as well? So what's your -- what's the
7 step after this project, in other words?
8 DR. BROOKS: So I guess I'm trying to be clear
9 because I'm talking about two different things. The
10 PaVe-GT is one and then the individualized therapies
11 that we're working with FNIH and FDA CBER -- is that
12 what you're referring to?
13 MR. THAKUR: No. You had a slide where you
14 talked about on one side you had here's what the FDA is
15 gonna review in great detail, and then ultimately, when
16 you bring on a new thing, it'll happen faster. And I'm
17 -- I think that was the PaVe-GT ultimate thinking. But
18 I'm not clear on what the final status that you're
19 trying to drive to, how you see the manufacturing, and
20 the FDA, and the NIH all working together.
21 DR. BROOKS: Yeah. I think there's different
www.transcriptionetc.com
361
1 levels and different projects. But I think the idea in
2 part was to make the regulatory process easier for the
3 FDA. When we're adding on -- if they want to add on --
4 we want to add on a new disease, if the vector
5 manufacturing is one they've seen before and the
6 biodistribution has been seen before that we wouldn't
7 have to repeat that. And that could increase our
8 clinical trial startup and make the regulatory path
9 easier.
10 I don't think I would imply that we're gonna
11 get FDA to approve a manufacturing process. I don't
12 see -- it's not obvious to me how that would work, but
13 I think it would be something like having a -- you
14 know, using the same process over and over again and
15 having the focus just be on what's different.
16 MR. THAKUR: And that'll be the clinical
17 center doing that going forward in long term?
18 DR. BROOKS: No. I don't think in long term.
19 I think optimally in long term we'd want to expand
20 this. And I think expanding beyond the PaVe-GT, our
21 pilot project, would ultimately hope- -- potentially be
www.transcriptionetc.com
362
1 the FNIH public-private partnership, and that would not
2 be limited to the NIH Clinical Center.
3 MR. THAKUR: Thank you.
4 DR. WITTEN: Before we take the next question,
5 I just want to clarify. So I -- what P.J. said is
6 correct. I mean, that is a shared, you know, his
7 description of it. We don't license processes. But if
8 we learn from the process or what happens and we learn
9 from our review and we learn from the science, that'll
10 facilitate continuing development, which is I think
11 what the goal of the program is. So that's -- I think
12 we're in agreement about what we think will happen, the
13 benefit could be. Yes?
14 MR. HORGAN: Rich Horgan from Cure Rare
15 Disease. One of the things tying together the ethical
16 and the stakeholder issues I think we may have been
17 overlooking a bit is the role of the payer in this. So
18 in the last two months we've had conversations with
19 chief medical officers of two of the biggest payers or
20 insurance companies in the United States. They are
21 aware of the development of customized therapeutics.
www.transcriptionetc.com
363
1 And the current mechanism for reimbursement is one that
2 is not at all conducive to reimbursing customized
3 therapies. More of a comment than a question, but an
4 urge to consider the payer perspective as you're
5 designing these, whether it's a platform trial or other
6 thinking both at the NIH as well as the FDA, because I
7 think, at the end of the day, if we can prove that we
8 have efficacious and safe custom drugs for one or two
9 patients, that's certainly great. But it's not
10 sustainable if we don't have payers on our side and
11 supportive of this approach.
12 So sort of urge thinking and more thought
13 surrounding that area because these certainly aren't
14 cheap, especially when we get to larger volume AAV
15 deliveries like with a Duchenne or another
16 neuromuscular disease. It's not an eye, and it's not
17 as privileged as the CNS in being compartmentalized to
18 some degree, but, you know, would urge some thinking
19 around that area.
20 DR. WITTEN: Thank you. Are there comments
21 from the panel on this topic?
www.transcriptionetc.com
364
1 DR. BATEMAN-HOUSE: I want to thank you for
2 bringing that up. I've been thinking a lot about
3 payers today. And I don't remember who said it, so I'm
4 not calling out names. But someone today said
5 something about, you know, well, if we had an add on
6 therapy that was safe, and even it was only a little
7 effective, why wouldn't we do it? And I immediately
8 thought because payers won't pay for it. So I think
9 you can't lose sight of that, especially when you're
10 thinking about access downstream.
11 So it's one thing to say in this interim
12 period, we don't need to worry about payers. But
13 whenever I talk with a company, and in this case
14 whenever I talk with an academic center or anybody
15 doing novel development, I would say think downstream.
16 Who you plan on using this product, and what evidence
17 do you need to get? At what level of certainty do you
18 need to convince payers to make that actually happen?
19 Because it's one thing to get FDA approval; it's
20 another thing to go through that other set of
21 gatekeepers which are payers.
www.transcriptionetc.com
365
1 DR. WITTEN: Thank you. I think --
2 DR. BROOKS: Oh --
3 DR. WITTEN: Sorry.
4 DR. BROOKS: I just -- I'll just make one
5 point. I think, I mean, I certainly agree about the
6 payer point. I think one of the other efforts that we
7 have in the Office of Rare Diseases Research is to try
8 to understand the cost of all these rare diseases on
9 our current health care system. Because when the
10 payers are going to be thinking about this, it's the
11 cost of paying for the therapy compared to the cost of
12 not having the therapy. And understanding the current
13 costs of all these rare diseases on our medical system
14 is -- we don't really have good data on that for a
15 variety of reasons, in part because of the difficulty
16 of the lack of ICD codes for some of these diseases.
17 So we have quite an effort going on at NCATS to come up
18 with a good estimate of what we call the cost of rare
19 because that will help the payer consideration.
20 DR. WITTEN: Do we have a question from the
21 online viewers?
www.transcriptionetc.com
366
1 THE OPERATOR: Yes, we do. The topic is
2 ethics.
3 DR. WITTEN: It's hard to hear you.
4 THE OPERATOR: The topic is ethics. And the
5 question is, in what I'm hearing, clinical trials are
6 very much being spoken of as treatment. How much
7 concern is there about research subjects or patients
8 clearly understanding and giving consent to early
9 trials that have not yet established safety?
10 DR. BATEMAN-HOUSE: So I think that was the
11 point that I was trying to make is that, traditionally,
12 we have said that there is research that should not be
13 thought of as therapy and that there is a high bar that
14 is expected to be cleared in terms of the informed
15 consent process that is asked of when a patient goes to
16 participate versus we had a much lower informed consent
17 process to participate in a therapeutic endeavor. And
18 the example I gave was surgery. So you know, you sign
19 a one page very small consent form to have your gall
20 bladder removed versus a very complex consent form to
21 participate in a clinical trial.
www.transcriptionetc.com
367
1 That's the model that we all grew up with.
2 That's the model that has always been said is a best
3 practice. But the question is are we at a point where
4 that model need transforming? And, in some cases, it
5 may be that it does need transforming.
6 If a truly bespoke therapy is being done on
7 one person even though it is experimental, it's not
8 necessarily research anymore. So we need to come up
9 with some understanding of how to navigate that divide.
10 And then the other point that I had hoped to make is
11 that that doesn't mean that that distinction is gonna
12 collapse across the board. So there is still going to
13 be areas where there is a divide between research and
14 therapy, and we still need to make sure that patients
15 in those contexts understand that going into this
16 clinical trial is not necessarily a therapeutic
17 endeavor.
18 And so there's a possibility of having mixed
19 messages about, you know, yay, all research is
20 treatment when that's really not true. And how do you
21 make those clarifications clear to potential research
www.transcriptionetc.com
368
1 subjects? I'm sorry. I'm looking at you because
2 you're the one that asked the question. I should be
3 looking -- I don't know who I'm supposed to be looking
4 at. The camera. Hello, camera.
5 So I just -- I'm very concerned in terms of
6 understanding, transparency, and expectations that we
7 need to be very clear about what is status quo, what is
8 different, and, if there are differences, why there are
9 differences and, if people are being held to the status
10 quo per se, why that is as opposed to it just being,
11 like, we like this disease and we don't like this
12 disease, or this disease has more engaged patients
13 versus this disease doesn't? I think we need to be
14 more clear as to why we're acting in certain ways in
15 certain paradigms.
16 DR. WITTEN: Thank you. I'm going to take the
17 two questions from the people who are already lined up
18 and then take the chair's prerogative to ask the final
19 question before closing the session.
20 MS. NOSRATIEH: Thank you. This is Anita
21 Nosratieh from FasterCures. This is a question going
www.transcriptionetc.com
369
1 back to the payer considerations and thinking about
2 this truly end-to-end access. Do you think, P.J., it
3 would be possible to incorporate CMS into the pilot
4 that you guys are spearheading between NIH, FDA? Just
5 seems like a natural, kind of, inclusion.
6 DR. BROOKS: I’m looking over at Peter. Okay.
7 Yeah. So Peter's going to address that later. Thank
8 you.
9 MS. BLACK: Hello. Lauren Black with Charles
10 River Laboratories and ex-CBER. I'm interested to see
11 the analogies between the current personalized medicine
12 and where we stand today in terms of monogenic
13 diseases. Within the context of monogenic diseases, I
14 think it's more like surgery, as Alison pointed out,
15 where the patient comes in. You can do an analysis
16 that's equivalent to saying, okay, the artery is
17 bleeding. We know what's wrong with the patient.
18 We know that they need a specific enzyme to be
19 replaced, or they need a certain gene replaced or
20 knocked out. We know exactly what's wrong with those
21 patients. So for that subset, this is very surgical.
www.transcriptionetc.com
370
1 It seems as if the payer would take a more surg- --
2 investigational surgery type approach to saying okay,
3 we're gonna replace that gene. We can measure that.
4 We can say if the initial drugs are working and have a
5 pharmacodynamic response.
6 And then say that yes, there's a blurred line
7 between treatment and research, but we can see that the
8 gene that was missing is now producing that protein.
9 And we can detect that protein in the blood and the
10 CSF. That seems to be a lot more clear than trying to
11 treat a disease that we didn't understand the cause or
12 had multiplicities of causes.
13 Here we have a much more specific thing that
14 we're looking to accomplish, so why shouldn't the lines
15 be blurred? Because as soon as we can detect the
16 replaced protein in the person, we know that we're much
17 closer to actually remediating their condition. So I
18 think this is actually a place where we can make a sea
19 change because we can see what's wrong, and we can see
20 how to fix it.
21 DR. WITTEN: Well, I think it's -- we're all
www.transcriptionetc.com
371
1 optimistic that this approach will work. So I think
2 we'll have to see what happens with this.
3 DR. BATEMAN-HOUSE: Can I say something about
4 it?
5 DR. WITTEN: Oh, sure.
6 DR. BATEMAN-HOUSE: So I'm not sure there was
7 a question there, and, if there was, I didn't get it.
8 So sorry if I don't answer correctly. But I just
9 wanted to say -- one thing that I just want to make
10 sure is clear, when I am saying that the intention may
11 be therapeutic but something being done is still
12 experimental and hence we need to figure out how to
13 deal with the informed consent and other problems of
14 that nature, is I want to share a conversation I had
15 with Dr. Timothy Yu who has been mentioned several
16 times.
17 So he had a terminally ill child who had a de
18 novo mutation of Batten's disease and was able to say,
19 I think I can come up with a customized therapy that
20 will help this child based on a platform from Spinraza
21 that I can make some alterations and use it to
www.transcriptionetc.com
372
1 potentially help this child. The thing that he was
2 concerned about he told me going into this was no one
3 had done this before, and there was a possibility that
4 by infusing this experimental product into this child,
5 who was blind, was not really able to communicate, and
6 was obviously headed towards an early demise, he could,
7 yes, actually, intervene in that trajectory, but he
8 could have other potentially unintended side effects.
9 And the one that he was concerned about was
10 awakening or reinvigorating some part of the brain that
11 would allow her to experience pain and thought I really
12 don’t want to give this intervention to a child that
13 may, yes, prolong her life but may also make her
14 current state of being worse. So I think even though
15 there's a very sort of cut and paste mentality, like,
16 of course this is gonna work -- it's very logical and
17 it could work -- it's still experimental. We don't
18 know what's gonna happen.
19 And even though we could say it's more of an
20 experimental therapy than research because the intent
21 is to help this person, you can't lose fact of the --
www.transcriptionetc.com
373
1 sight of the fact that it is experimental. And you
2 have to have that understanding going into it and make
3 sure that there is the informed consent and make sure
4 that everybody understands, you now, we don't know
5 what's gonna happen here. Maybe after a couple
6 iterations in a couple different settings, we'll have a
7 better basis for being able to make predictions but not
8 at first.
9 DR. WITTEN: Thank you. So my last question --
10 and I'm gonna ask everyone on the panel. And I'll
11 start with Jill, if that's okay -- is -- of course you
12 don't know the question, so I guess it's maybe not fair
13 to ask if it's okay. Is, if there's one thing -- so we
14 obviously at CBER are looking at these questions very
15 carefully as to what we can do to facilitate the
16 process or what we need to look at, what our next steps
17 should be to try to benefit patients and benefit
18 product development.
19 And so I just would like to know from each of
20 you, if you have any thoughts, if you do have any
21 thoughts on what would success look like for us just
www.transcriptionetc.com
374
1 for the next year? I don't mean 10 years, you know, 20
2 years success, but just, if we could accomplish
3 something in the next year, what would it look like?
4 If you have a comment on that.
5 MS. WOOD: Well, I think what P.J. is
6 suggesting is profound, and it's absolutely amazing.
7 And if you could pull it off, you'd be pulling off a
8 decade of success right there. Something very easy I
9 would say, a success is to identify those ultra-rare
10 diseases and uber rare diseases that really need --
11 have a need for mouse models and registries and natural
12 history studies and help facilitate those --
13 development of those things by either working with
14 their patient organizations or trying to figure out an
15 in-house way of doing that for all.
16 DR. WITTEN: Thank you. Next. Alison.
17 DR. BATEMAN-HOUSE: I think the fairest answer
18 to say is I don't know.
19 DR. WITTEN: Okay.
20 DR. BATEMAN-HOUSE: But I guess I would just
21 say I really think it's important to be transparent in
www.transcriptionetc.com
375
1 these ongoing conversations and also, in terms of -- to
2 the extent that CBER is starting to lean certain ways,
3 to divulge that as soon as possible because it sounds
4 like there's a lot of people waiting for some sort of -
5 - certainly as to whether they're on the right path or
6 not. And the sooner that they can feel some sense of
7 assurance that, you know, maybe we don't have a final
8 guidance yet, but we see that there's a wind blowing
9 this way, that would be helpful.
10 DR. WITTEN: Thank you.
11 DR. BROOKS: So I think one of the most
12 exciting things that I see is the effort that involved
13 with Peter and FNIH to develop this public-private
14 partnership and really test a very different way to do
15 gene therapy clinical trials for bespoke therapies. I
16 don't know that we can pull it off and get it started
17 within a year. But I think if we can do that, to have
18 the FDA leadership involved in an effort like this
19 seems like a very different approach for the FDA
20 leadership. And I think it's really wonderful and
21 exciting for all of us who are participating in it and
www.transcriptionetc.com
376
1 I think, you know, for the whole community. So…
2 DR. WITTEN: Thank you.
3 DR. SCHOOLEY: You know, we're working with
4 often rare diseases. And we should try to learn how to
5 generalize our knowledge base, so we don't have to
6 discover the same thing over and over again in each
7 specific clinical indication, so learning how to
8 generalize with the skepticism you need about over
9 generalization, at the same time -- and to focus our
10 resources on moving the field forward rather than just
11 repetitively doing the same thing over and over again,
12 just say this is the way we develop this therapeutic.
13 In other words, connecting the dots in a more -- among
14 these efforts in a more cross-fertilizing way will help
15 us move forward. And I think that, at the end of the
16 day, if we can do that, we'll help in both these kinds
17 of diseases and how we approach other diseases as well.
18 The innovation that we've seen today in these
19 approaches, I think, is really breathtaking, and that
20 innovation needs to be balanced with the care that we
21 approach human engagement in research.
www.transcriptionetc.com
377
1 But also, we need to make sure that what we
2 learn on platform that has a lot of similarities to
3 others is shared so we didn't have to discover the same
4 thing over and over again. And parallel efforts, that
5 holds everybody back because of proprietary needs that
6 put these people at risk over and over again who
7 studied the same thing and slows the field. So
8 generalization early in the process helps everyone,
9 particularly our patients, so early sharing of
10 approaches, platforms, techniques. We're all here for
11 the same thing, and there's plenty of room for
12 innovation in any given field is what I would argue.
13 DR. WITTEN: Thank you. And as usual FDA gets
14 the last word, so…
15 CAPT. VAILLANCOURT: Thank you. Well, I think
16 today is an extremely important day. We're starting a
17 public dialogue about this critically important area
18 and in a way of -- to go forward without having all the
19 burden on patients and parents, such as Jill and others
20 out there. And I think it's so important that we keep
21 the dialogue going, that we don't lose momentum.
www.transcriptionetc.com
378
1 And I'd like to agree with what P.J. said,
2 and, being from FDA, I think this announcement of this
3 public-private partnership is very, very exciting. So
4 be great if we could convene again and see some -- hear
5 about the status of what's happening with that public-
6 private partnership, also to get an update on how the
7 NIH program is going. I mean, these are all really
8 exciting initiatives -- but also to hear more from our
9 stakeholders.
10 We're just so thrilled that everybody is here.
11 For those of you in the room and everybody on the
12 phone, the whole intention was to get the stakeholders
13 together today and to hear from everyone. So again,
14 keep the dialogue going, keep the momentum going.
15 DR. WITTEN: Thank you. I'd like to thank the
16 panel and the speakers. And next I'm going to turn it
17 over to Dr. Peter Marks, our Center Director for -- to
18 wrap it up.
19
20 WRAP UP AND CLOSING REMARKS: DR. PETER MARKS
21
www.transcriptionetc.com
379
1 DR. MARKS: So thank you everyone in the room
2 and online who had stuck with us for the full day. I
3 think rather than summarizing each of the sessions, I
4 think what I just want to say is, I think what really
5 came through pretty clearly is that I think we all see
6 the compelling need to make headway here in these
7 individualized or bespoke therapies. And I think we --
8 in each of the sessions we have these building blocks
9 that we can build upon, whether it's on the
10 manufacturing, the non-clinical aspects, the clinical
11 aspects, or patient access to these things.
12 Just to back up to try to address some of this
13 because the issue of how do you pay for these things
14 has come up. So I think there are lots of different
15 ways to think about this. But the way I would think
16 about this for at least for the reason why we're very
17 interested in a public-private partnership is I think
18 we're in a time of transition here. We're in a time of
19 transition where -- there was a model several years ago
20 where every rare disease would be commercially viable
21 somehow because you could charge enough for the gene
www.transcriptionetc.com
380
1 therapy.
2 I think it's -- I'm oversimplifying it, but I
3 think what we've realized is -- and what companies have
4 realized is that commercial viability lies beyond many
5 of the diseases that we're talking about today, which
6 means we have to find some other way to fill in that
7 gap. I will tell you an opinion. This is not the
8 opinion of the United States Government Health, and
9 Human Services or the Food and Drug Administration of
10 the United States of America.
11 This is my own opinion that, 10 to 15 years
12 from now, this issue will be fixed because there will
13 be commercial viability for very rare bespoke therapies
14 inasmuch as I think much of this will be essentially
15 dealt with by having machines that can -- this will be
16 a device issue. Many of these gene therapies will be
17 made potentially on non-viral platforms or by
18 mechanisms that don't require the kind of setup that
19 they currently do. And we will have had a lot more
20 experience about what you can leverage.
21 But for this interval of the next 5 to 10
www.transcriptionetc.com
381
1 years, I think we have to find a way to get these
2 therapies to individuals in need. I don't know that
3 we're gonna find a way to get payers to pay for them,
4 at least in the short term. But what I do think is
5 there's a lot of good will going between companies,
6 non-profit organizations, and, for that matter,
7 government wanting to collaborate with them to find
8 ways to try to make these therapies available to those
9 in need.
10 There's gonna be a lot of ethical issues, a
11 lot of prioritization issues that'll have to be worked
12 through in this. But I think that's the goal of these
13 public-private partnerships is to try to find a way
14 forward. And ultimately, the reason why this is so
15 important, at least to me, is that this is a case
16 where, if we can get it right for these very small
17 numbers of patients, ultimately the entire field of
18 gene therapy is bound to benefit. So it's one of these
19 things that start small and local and then go more
20 globally.
21 And just so that I just mention that we didn't
www.transcriptionetc.com
382
1 concentrate here today on the discussion of the more
2 global issues for gene therapy. But part of the reason
3 for getting it right here is that, if we can take care
4 of these products on a small scale here, hopefully, we
5 can have a global framework so that patients around the
6 globe will benefit from their development. It really
7 would be a shame if we spent the time developing these
8 here, and then they're not accessible -- you know, it
9 would be really sad for a Sanfilippo type C or a type
10 III C or D patient here not to get something and not to
11 have a patient in Asia or in Africa benefit from that
12 same advance that we've made.
13 So ultimately, I think we're gonna go back
14 from this. We will do what FDA likes to do is we will
15 cogitate for a while. I think we do understand that
16 ultimately the way we try to put forth our thinking is
17 in guidance. Hopefully that will be forthcoming at
18 some point in the future and in the not too distant
19 future.
20 And we'll also continue to work with our
21 partners at FNIH and NCATS to try to move forward this
www.transcriptionetc.com
383
1 public-private partnership. Is success assured? No,
2 it's not. But I think it's certainly worth a try
3 because there's a lot of good will there. There are a
4 lot of patients in need of these therapies, and I do
5 think we have to try to do something differently that
6 will try to get us there.
7 We have actually -- just to answer another
8 question, we have actually spent some time with
9 business folks and with companies talking about
10 economies of scale. And there probably are some
11 economies of scale to be had here in part by using
12 excess capacity, in part by reusing certain aspects of
13 files and et cetera. It's not gonna ever be cheap, but
14 we do think this is something that could hopefully lead
15 us to be able to more efficiently get there for
16 patients.
17 So with that I just want to close by saying I
18 really want to thank all of the speakers today,
19 particularly Ms. Wood who I really think really shared
20 a very compelling picture of what it's like to deal
21 with this type of situation from a variety of aspects.
www.transcriptionetc.com
384
1 And to all of you for -- thank you for coming today and
2 for really caring about this issue. And I'd also echo
3 something that Julie Vaillancourt said which is that we
4 look forward to continuing the dialogue with everyone.
5 And with that I have two last things to do.
6 One of them is to once again to thank Leslie Haynes and
7 Gopa Raychaudhuri for really planning an incredibly
8 excellent workshop. So let's give them a round of
9 applause. And then I told Gopa I'd give her the last
10 word. So here she goes. You're good. Okay. With
11 that, thank you very much. Okay. Thanks again,
12 everyone.
13 [MEETING ADJOURNED]
www.transcriptionetc.com